{"object":"list","data":[{"id":"claude-haiku-5.5","object":"model","created":1791446400,"owned_by":"Anthropic","description":"Claude Haiku 5.5 is Anthropic’s fastest and most cost-efficient small model, built for high-volume workloads that require responsive performance, low operating costs, and reliable handling of repetitive tasks and lightweight agentic workflows.","pricing":{"currency":"EUR","input_token":0.101,"output_token":0.505,"cache_read_cost":0.01,"cache_write_cost":0.126},"providers":["amazon_ireland","amazon_paris","google"],"context_size":1000000,"max_output_tokens":128000,"tags":["Instruct","Tools","Image","Reasoning","Document"],"input_modalities":["text","image","file"],"output_modalities":["text"],"supported_features":["json_mode","reasoning","tools"],"model_series":["claude-haiku-5"],"providers_details":{"amazon_ireland":{"pricing":{"currency":"EUR","input_token":0.101,"output_token":0.505,"cache_read_cost":0.01,"cache_write_cost":0.126},"quantization":null,"context_size":1000000,"max_output_tokens":128000,"supported_features":["json_mode","reasoning","tools"],"metrics":{"throughput":664.3,"latency":6.38,"ttft":3.73,"cache_hit_rate":0.32,"uptime":1.0}},"amazon_paris":{"pricing":{"currency":"EUR","input_token":0.101,"output_token":0.505,"cache_read_cost":0.01,"cache_write_cost":0.126},"quantization":null,"context_size":1000000,"max_output_tokens":128000,"supported_features":["json_mode","reasoning","tools"],"metrics":{"throughput":659.89,"latency":5.11,"ttft":3.97,"cache_hit_rate":0.46,"uptime":1.0}},"google":{"pricing":{"currency":"EUR","input_token":0.101,"output_token":0.505,"cache_read_cost":0.01,"cache_write_cost":0.126},"quantization":null,"context_size":1000000,"max_output_tokens":128000,"supported_features":["json_mode","reasoning","tools"],"metrics":{"throughput":337.64,"latency":1.74,"ttft":1.54,"cache_hit_rate":0.32,"uptime":1.0}}}},{"id":"mistral-large-4-0","object":"model","created":1791280393,"owned_by":"Mistral AI","description":"Mistral Large 4 is a trillion-parameter multimodal MoE model built for frontier-level reasoning, enterprise workloads, and multilingual applications, with 49B active parameters per token.","pricing":{"currency":"EUR","input_token":0.687,"output_token":2.11,"cache_read_cost":0.071},"providers":["mistral"],"context_size":1000000,"max_output_tokens":262144,"tags":["Instruct","Tools","Image","Reasoning"],"input_modalities":["text","image"],"output_modalities":["text"],"supported_features":["json_mode","reasoning","tools"],"model_series":["mistral-large"],"providers_details":{"mistral":{"pricing":{"currency":"EUR","input_token":0.687,"output_token":2.11,"cache_read_cost":0.071},"quantization":null,"context_size":1000000,"max_output_tokens":262144,"supported_features":["json_mode","reasoning","tools"],"metrics":{"throughput":91.88,"latency":17.91,"ttft":3.09,"cache_hit_rate":0.86,"uptime":1.0}}}},{"id":"claude-sonnet-5.5","object":"model","created":1791275882,"owned_by":"Anthropic","description":"Claude Sonnet 5.5 is Anthropic’s mid-tier AI model, designed to deliver near-flagship coding and knowledge-work performance with faster execution, lower token usage, and reduced cost for everyday professional tasks.","pricing":{"currency":"EUR","input_token":2.02,"output_token":10.098,"cache_read_cost":0.202,"cache_write_cost":2.525},"providers":["google","amazon_ireland","amazon_paris"],"context_size":1000000,"max_output_tokens":128000,"tags":["Instruct","Tools","Image","Reasoning","Document"],"input_modalities":["text","image","file"],"output_modalities":["text"],"supported_features":["json_mode","reasoning","tools"],"model_series":["claude-sonnet-5"],"providers_details":{"google":{"pricing":{"currency":"EUR","input_token":2.02,"output_token":10.098,"cache_read_cost":0.202,"cache_write_cost":2.525},"quantization":null,"context_size":1000000,"max_output_tokens":128000,"supported_features":["json_mode","reasoning","tools"],"metrics":{"throughput":265.19,"latency":9.96,"ttft":14.5,"cache_hit_rate":0.69,"uptime":0.99}},"amazon_ireland":{"pricing":{"currency":"EUR","input_token":2.02,"output_token":10.098,"cache_read_cost":0.202,"cache_write_cost":2.525},"quantization":null,"context_size":1000000,"max_output_tokens":128000,"supported_features":["json_mode","reasoning","tools"],"metrics":{"throughput":312.65,"latency":8.64,"ttft":7.58,"cache_hit_rate":0.37,"uptime":0.94}},"amazon_paris":{"pricing":{"currency":"EUR","input_token":2.02,"output_token":10.098,"cache_read_cost":0.202,"cache_write_cost":2.525},"quantization":null,"context_size":1000000,"max_output_tokens":128000,"supported_features":["json_mode","reasoning","tools"],"metrics":{"throughput":343.15,"latency":7.78,"ttft":7.1,"cache_hit_rate":0.82,"uptime":0.97}}}},{"id":"gpt-6.1-sol","object":"model","created":1790683200,"owned_by":"OpenAI","description":"GPT-6.1 Sol is a high-efficiency mid-tier model delivering near-flagship coding, reasoning, and agentic performance at significantly lower cost.","pricing":{"currency":"EUR","input_token":2.154,"output_token":10.771,"cache_read_cost":0.108,"cache_write_cost":2.693},"providers":["azure_spc"],"context_size":1050000,"max_output_tokens":128000,"tags":["Instruct","Reasoning","Tools","Image","Document"],"input_modalities":["text","image","file"],"output_modalities":["text"],"supported_features":["json_mode","reasoning","tools"],"model_series":["gpt-6-sol"],"providers_details":{"azure_spc":{"pricing":{"currency":"EUR","input_token":2.154,"output_token":10.771,"cache_read_cost":0.108,"cache_write_cost":2.693},"quantization":null,"context_size":1050000,"max_output_tokens":128000,"supported_features":["json_mode","reasoning","tools"],"metrics":{"throughput":73.66,"latency":9.46,"ttft":2.82,"cache_hit_rate":0.9,"uptime":1.0}}}},{"id":"gpt-6-sol","object":"model","created":1790164800,"owned_by":"OpenAI","description":"GPT-6 Sol is a balanced mid-tier model offering strong coding, data analysis, and multi-step reasoning capabilities.","pricing":{"currency":"EUR","input_token":2.154,"output_token":10.771,"cache_read_cost":0.215,"cache_write_cost":2.693},"providers":["azure_spc"],"context_size":1050000,"max_output_tokens":128000,"tags":["Instruct","Reasoning","Tools","Image","Document"],"input_modalities":["text","image","file"],"output_modalities":["text"],"supported_features":["json_mode","reasoning","tools"],"model_series":["gpt-6-sol","gpt-6"],"providers_details":{"azure_spc":{"pricing":{"currency":"EUR","input_token":2.154,"output_token":10.771,"cache_read_cost":0.215,"cache_write_cost":2.693},"quantization":null,"context_size":1050000,"max_output_tokens":128000,"supported_features":["json_mode","reasoning","tools"],"metrics":{"throughput":189.76,"latency":7.55,"ttft":2.1,"cache_hit_rate":0.3,"uptime":1.0}}}},{"id":"gpt-6-luna","object":"model","created":1790161200,"owned_by":"OpenAI","description":"GPT-6 Luna is a high-speed, cost-efficient model in OpenAI’s GPT-6 family, designed for large-scale, latency-sensitive workloads.","pricing":{"currency":"EUR","input_token":0.108,"output_token":0.539,"cache_read_cost":0.011,"cache_write_cost":0.135},"providers":["azure_spc"],"context_size":1050000,"max_output_tokens":128000,"tags":["Instruct","Reasoning","Tools","Image","Document"],"input_modalities":["text","image","file"],"output_modalities":["text"],"supported_features":["json_mode","reasoning","tools"],"model_series":["gpt-6-luna","gpt-6"],"providers_details":{"azure_spc":{"pricing":{"currency":"EUR","input_token":0.108,"output_token":0.539,"cache_read_cost":0.011,"cache_write_cost":0.135},"quantization":null,"context_size":1050000,"max_output_tokens":128000,"supported_features":["json_mode","reasoning","tools"],"metrics":{"throughput":233.9,"latency":3.19,"ttft":5.73,"cache_hit_rate":0.86,"uptime":0.98}}}},{"id":"claude-opus-5.5","object":"model","created":1790145000,"owned_by":"Anthropic","description":"Claude Opus 5.5 is Anthropic’s most capable Opus model, built for high-end coding and long-running agent workflows.","pricing":{"currency":"EUR","input_token":3.949,"output_token":19.747,"cache_read_cost":0.197,"cache_write_cost":4.937},"providers":["amazon_ireland","amazon_paris","google"],"context_size":1000000,"max_output_tokens":128000,"tags":["Instruct","Tools","Image","Reasoning","Document"],"input_modalities":["text","image","file"],"output_modalities":["text"],"supported_features":["json_mode","reasoning","tools"],"model_series":["claude-opus-5"],"providers_details":{"amazon_ireland":{"pricing":{"currency":"EUR","input_token":3.949,"output_token":19.747,"cache_read_cost":0.197,"cache_write_cost":4.937},"quantization":null,"context_size":1000000,"max_output_tokens":128000,"supported_features":["json_mode","reasoning","tools"],"metrics":{"throughput":238.34,"latency":12.23,"ttft":12.23,"cache_hit_rate":0.6,"uptime":0.96}},"amazon_paris":{"pricing":{"currency":"EUR","input_token":3.949,"output_token":19.747,"cache_read_cost":0.197,"cache_write_cost":4.937},"quantization":null,"context_size":1000000,"max_output_tokens":128000,"supported_features":["json_mode","reasoning","tools"],"metrics":{"throughput":225.74,"latency":11.33,"ttft":9.44,"cache_hit_rate":0.77,"uptime":0.96}},"google":{"pricing":{"currency":"EUR","input_token":3.949,"output_token":19.747,"cache_read_cost":0.197,"cache_write_cost":4.937},"quantization":null,"context_size":1000000,"max_output_tokens":128000,"supported_features":["json_mode","reasoning","tools"],"metrics":{"throughput":283.71,"latency":13.44,"ttft":7.8,"cache_hit_rate":0.5,"uptime":0.99}}}},{"id":"deepseek-v4.1-flash","object":"model","created":1789372111,"owned_by":"Deepseek","description":"DeepSeek V4.1 Flash is an efficient multimodal MoE model optimized for fast agentic reasoning and coding workflows.","pricing":{"currency":"EUR","input_token":0.18,"output_token":0.539,"cache_read_cost":0.004},"providers":["tensorix","arkane","aki","entrim","sference"],"context_size":1048576,"max_output_tokens":1048576,"tags":["Instruct","Tools","Reasoning","Image"],"input_modalities":["text","image"],"output_modalities":["text"],"supported_features":["json_mode","reasoning","tools"],"model_series":["deepseek-v4-flash"],"providers_details":{"tensorix":{"pricing":{"currency":"EUR","input_token":0.449,"output_token":1.346,"cache_read_cost":0.117},"quantization":"fp8","context_size":1048576,"max_output_tokens":64000,"supported_features":["json_mode","reasoning","tools"],"metrics":{"throughput":345.6,"latency":3.48,"ttft":2.08,"cache_hit_rate":0.54,"uptime":1.0}},"arkane":{"pricing":{"currency":"EUR","input_token":0.269,"output_token":1.077,"cache_read_cost":0.067},"quantization":null,"context_size":1048576,"max_output_tokens":393216,"supported_features":["json_mode","reasoning","tools"],"metrics":{"throughput":146.58,"latency":7.61,"ttft":3.28,"cache_hit_rate":0.84,"uptime":0.99}},"aki":{"pricing":{"currency":"EUR","input_token":0.25,"output_token":1.0,"cache_read_cost":0.1},"quantization":"fp8","context_size":1048576,"max_output_tokens":81920,"supported_features":["json_mode","reasoning","tools"],"metrics":{"throughput":139.97,"latency":5.93,"ttft":5.53,"cache_hit_rate":0.86,"uptime":0.99}},"entrim":{"pricing":{"currency":"EUR","input_token":0.18,"output_token":0.539,"cache_read_cost":0.004},"quantization":"fp8","context_size":1048576,"max_output_tokens":393216,"supported_features":["json_mode","reasoning","tools"],"metrics":{"throughput":167.45,"latency":14.51,"ttft":1.68,"cache_hit_rate":0.94,"uptime":1.0}},"sference":{"pricing":{"currency":"EUR","input_token":0.444,"output_token":1.332,"cache_read_cost":0.044},"quantization":"nvfp4","context_size":1048576,"max_output_tokens":1048576,"supported_features":["json_mode","reasoning","tools"],"metrics":{"throughput":483.14,"latency":2.98,"ttft":0.62,"cache_hit_rate":0.9,"uptime":0.99}}}},{"id":"gemini-3.8-flash","object":"model","created":1788853082,"owned_by":"Google","description":"Gemini 3.8 Flash is a fast, cost-efficient multimodal model built for advanced coding and agentic workflows.","pricing":{"currency":"EUR","input_token":0.741,"output_token":3.703,"audio_cost":1.851e-05,"cache_read_cost":0.074,"cache_write_cost":0.075},"providers":["google","google_priority"],"context_size":1048576,"max_output_tokens":65535,"tags":["Instruct","Tools","Image","Reasoning","Audio","Document"],"input_modalities":["text","image","audio","file"],"output_modalities":["text"],"supported_features":["json_mode","reasoning","tools"],"model_series":["gemini-3-flash"],"providers_details":{"google":{"pricing":{"currency":"EUR","input_token":0.741,"output_token":3.703,"audio_cost":1.851e-05,"cache_read_cost":0.074,"cache_write_cost":0.075},"quantization":null,"context_size":1048576,"max_output_tokens":65535,"supported_features":["json_mode","reasoning","tools"],"metrics":{"throughput":177.46,"latency":9.84,"ttft":8.14,"cache_hit_rate":0.37,"uptime":1.0}},"google_priority":{"pricing":{"currency":"EUR","input_token":1.333,"output_token":6.665,"audio_cost":3.332e-05,"cache_read_cost":0.133,"cache_write_cost":0.898},"quantization":null,"context_size":1048576,"max_output_tokens":65535,"supported_features":["json_mode","reasoning","tools"],"metrics":{"throughput":195.72,"latency":9.93,"ttft":3.23,"cache_hit_rate":0.01,"uptime":1.0}}}},{"id":"deepseek-v4-pro-0813","object":"model","created":1788161882,"owned_by":"Deepseek","description":"DeepSeek V4 Pro 0813 is a flagship MoE model delivering efficient long-context reasoning, coding, and agentic performance.","pricing":{"currency":"EUR","input_token":1.795,"output_token":3.59,"cache_read_cost":0.449},"providers":["tensorix"],"context_size":1048576,"max_output_tokens":64000,"tags":["Instruct","Tools","Reasoning"],"input_modalities":["text"],"output_modalities":["text"],"supported_features":["json_mode","reasoning","tools"],"model_series":["deepseek-v4-pro"],"providers_details":{"tensorix":{"pricing":{"currency":"EUR","input_token":1.795,"output_token":3.59,"cache_read_cost":0.449},"quantization":"fp8","context_size":1048576,"max_output_tokens":64000,"supported_features":["json_mode","reasoning","tools"],"metrics":{"throughput":167.99,"latency":2.3,"ttft":1.78,"cache_hit_rate":0.95,"uptime":0.97}}}},{"id":"glm-5.3","object":"model","created":1787936282,"owned_by":"Z.ai","description":"GLM-5.3 is a flagship open-weight model optimized for advanced coding, agentic workflows, and long-context reasoning.","pricing":{"currency":"EUR","input_token":0.784,"output_token":2.968,"cache_read_cost":0.198},"providers":["tensorix","inceptron","aki","mistral","arkane","sference"],"context_size":1048576,"max_output_tokens":1048576,"tags":["Instruct","Reasoning","Tools"],"input_modalities":["text"],"output_modalities":["text"],"supported_features":["json_mode","reasoning","tools"],"model_series":["glm-5"],"providers_details":{"tensorix":{"pricing":{"currency":"EUR","input_token":1.571,"output_token":4.039,"cache_read_cost":0.393},"quantization":"fp8","context_size":1048576,"max_output_tokens":64000,"supported_features":["json_mode","reasoning","tools"],"metrics":{"throughput":209.83,"latency":9.36,"ttft":3.16,"cache_hit_rate":0.71,"uptime":1.0}},"inceptron":{"pricing":{"currency":"EUR","input_token":0.784,"output_token":2.968,"cache_read_cost":0.198},"quantization":"fp4","context_size":1048576,"max_output_tokens":1048576,"supported_features":["json_mode","reasoning","tools"],"metrics":{"throughput":185.76,"latency":4.76,"ttft":1.25,"cache_hit_rate":0.92,"uptime":1.0}},"aki":{"pricing":{"currency":"EUR","input_token":1.0,"output_token":3.5,"cache_read_cost":0.25},"quantization":"nvfp4","context_size":512000,"max_output_tokens":81920,"supported_features":["json_mode","reasoning","tools"],"metrics":{"throughput":121.19,"latency":13.48,"ttft":4.92,"cache_hit_rate":0.83,"uptime":1.0}},"mistral":{"pricing":{"currency":"EUR","input_token":1.382,"output_token":4.344,"cache_read_cost":0.138},"quantization":null,"context_size":1000000,"max_output_tokens":128000,"supported_features":["json_mode","reasoning","tools"],"metrics":{"throughput":268.71,"latency":1.73,"ttft":1.75,"cache_hit_rate":0.93,"uptime":1.0}},"arkane":{"pricing":{"currency":"EUR","input_token":1.257,"output_token":3.949,"cache_read_cost":0.269},"quantization":null,"context_size":200000,"max_output_tokens":100000,"supported_features":["json_mode","reasoning","tools"],"metrics":{"throughput":129.68,"latency":18.93,"ttft":5.13,"cache_hit_rate":0.86,"uptime":1.0}},"sference":{"pricing":{"currency":"EUR","input_token":1.077,"output_token":3.77,"cache_read_cost":0.233},"quantization":"nvfp4","context_size":1048576,"max_output_tokens":1048576,"supported_features":["json_mode","reasoning","tools"],"metrics":{"throughput":209.82,"latency":3.59,"ttft":0.48,"cache_hit_rate":0.9,"uptime":1.0}}}},{"id":"qwen3.8-flash-next","object":"model","created":1787909762,"owned_by":"Alibaba Cloud","description":"Qwen3.8 Flash Next is an efficient multimodal MoE model designed for low-cost inference and strong coding performance.","pricing":{"currency":"EUR","input_token":0.18,"output_token":0.449,"cache_read_cost":0.045},"providers":["tensorix"],"context_size":262144,"max_output_tokens":64000,"tags":["Instruct","Reasoning","Tools","Image"],"input_modalities":["text","image"],"output_modalities":["text"],"supported_features":["json_mode","reasoning","tools"],"model_series":["qwen-3-flash"],"providers_details":{"tensorix":{"pricing":{"currency":"EUR","input_token":0.18,"output_token":0.449,"cache_read_cost":0.045},"quantization":"fp8","context_size":262144,"max_output_tokens":64000,"supported_features":["json_mode","reasoning","tools"],"metrics":{"throughput":195.96,"latency":7.03,"ttft":1.64,"cache_hit_rate":0.87,"uptime":0.99}}}},{"id":"glm-5.3-flash","object":"model","created":1787907482,"owned_by":"Z.ai","description":"GLM 5.3 Flash is a fast, cost-efficient multimodal model built for long-context reasoning and scalable workflows.","pricing":{"currency":"EUR","input_token":0.09,"output_token":0.314,"cache_read_cost":0.016},"providers":["tensorix","berget","entrim","inceptron","sference"],"context_size":1048576,"max_output_tokens":1048576,"tags":["Instruct","Reasoning","Tools","Image"],"input_modalities":["text","image"],"output_modalities":["text"],"supported_features":["json_mode","reasoning","tools"],"model_series":["glm-5-flash"],"providers_details":{"tensorix":{"pricing":{"currency":"EUR","input_token":0.18,"output_token":0.449,"cache_read_cost":0.045},"quantization":"fp8","context_size":1048576,"max_output_tokens":64000,"supported_features":["json_mode","reasoning","tools"],"metrics":{"throughput":206.84,"latency":10.78,"ttft":4.15,"cache_hit_rate":0.75,"uptime":1.0}},"berget":{"pricing":{"currency":"EUR","input_token":0.25,"output_token":0.5},"quantization":"fp8","context_size":1048576,"max_output_tokens":1048576,"supported_features":["json_mode","reasoning","tools"],"metrics":{"throughput":186.72,"latency":9.05,"ttft":3.35,"cache_hit_rate":0.89,"uptime":1.0}},"entrim":{"pricing":{"currency":"EUR","input_token":0.09,"output_token":0.314,"cache_read_cost":0.016},"quantization":"nvfp4","context_size":1048576,"max_output_tokens":131072,"supported_features":["json_mode","reasoning","tools"],"metrics":{"throughput":174.57,"latency":9.55,"ttft":2.45,"cache_hit_rate":0.94,"uptime":1.0}},"inceptron":{"pricing":{"currency":"EUR","input_token":0.135,"output_token":0.449,"cache_read_cost":0.063},"quantization":"fp8","context_size":1048576,"max_output_tokens":1048576,"supported_features":["json_mode","reasoning","tools"],"metrics":{"throughput":189.71,"latency":7.9,"ttft":1.37,"cache_hit_rate":0.86,"uptime":0.99}},"sference":{"pricing":{"currency":"EUR","input_token":0.18,"output_token":0.539,"cache_read_cost":0.063},"quantization":"nvfp4","context_size":1048576,"max_output_tokens":1048576,"supported_features":["json_mode","reasoning","tools"],"metrics":{"throughput":266.26,"latency":4.08,"ttft":1.09,"cache_hit_rate":0.84,"uptime":1.0}}}},{"id":"qwen3.8-27b","object":"model","created":1786966562,"owned_by":"Alibaba Cloud","description":"Qwen3.8 27B is a compact open-weight multimodal model optimized for efficient reasoning, coding, and agentic workflows on local hardware.","pricing":{"currency":"EUR","input_token":0.09,"output_token":0.359,"cache_read_cost":0.036},"providers":["tensorix","berget","aki","ionos","ovh","entrim","scaleway","arkane"],"context_size":1000000,"max_output_tokens":500000,"tags":["Instruct","Reasoning","Tools","Image"],"input_modalities":["text","image"],"output_modalities":["text"],"supported_features":["json_mode","reasoning","tools"],"model_series":["qwen-3-27b"],"providers_details":{"tensorix":{"pricing":{"currency":"EUR","input_token":0.359,"output_token":2.154,"cache_read_cost":0.09},"quantization":"fp8","context_size":262144,"max_output_tokens":262144,"supported_features":["json_mode","reasoning","tools"],"metrics":{"throughput":33.09,"latency":37.6,"ttft":6.22,"cache_hit_rate":0.27,"uptime":0.96}},"berget":{"pricing":{"currency":"EUR","input_token":0.4,"output_token":3.0},"quantization":"fp8","context_size":262000,"max_output_tokens":262000,"supported_features":["json_mode","reasoning","tools"],"metrics":{"throughput":34.92,"latency":385.58,"ttft":4.56,"cache_hit_rate":0.84,"uptime":1.0}},"aki":{"pricing":{"currency":"EUR","input_token":0.3,"output_token":2.2,"cache_read_cost":0.1},"quantization":"fp16","context_size":262144,"max_output_tokens":262144,"supported_features":["json_mode","reasoning","tools"],"metrics":{"throughput":59.51,"latency":199.64,"ttft":19.94,"cache_hit_rate":0.63,"uptime":1.0}},"ionos":{"pricing":{"currency":"EUR","input_token":0.4,"output_token":2.7},"quantization":"int4","context_size":262000,"max_output_tokens":262000,"supported_features":["json_mode","reasoning","tools"],"metrics":{"throughput":49.2,"latency":56.12,"ttft":35.12,"cache_hit_rate":null,"uptime":1.0}},"ovh":{"pricing":{"currency":"EUR","input_token":0.4,"output_token":2.7},"quantization":"fp8","context_size":262000,"max_output_tokens":262000,"supported_features":["json_mode","reasoning","tools"],"metrics":{"throughput":88.7,"latency":36.03,"ttft":20.81,"cache_hit_rate":null,"uptime":1.0}},"entrim":{"pricing":{"currency":"EUR","input_token":0.09,"output_token":0.359,"cache_read_cost":0.036},"quantization":"fp8","context_size":262144,"max_output_tokens":65536,"supported_features":["json_mode","reasoning","tools"],"metrics":{"throughput":82.86,"latency":18.96,"ttft":2.88,"cache_hit_rate":0.94,"uptime":0.99}},"scaleway":{"pricing":{"currency":"EUR","input_token":0.6,"output_token":3.3,"cache_read_cost":0.12},"quantization":null,"context_size":262000,"max_output_tokens":32768,"supported_features":["json_mode","reasoning","tools"],"metrics":{"throughput":169.11,"latency":9.82,"ttft":2.84,"cache_hit_rate":0.39,"uptime":1.0}},"arkane":{"pricing":{"currency":"EUR","input_token":0.09,"output_token":0.898},"quantization":null,"context_size":1000000,"max_output_tokens":500000,"supported_features":["json_mode","reasoning","tools"],"metrics":{"throughput":83.41,"latency":22.89,"ttft":9.27,"cache_hit_rate":0.22,"uptime":0.91}}}},{"id":"gemini-3.7-flash","object":"model","created":1786693082,"owned_by":"Google","description":"Gemini 3.7 Flash is a high-efficiency model delivering strong agentic capabilities, improved coding performance, and reliable multi-step execution.","pricing":{"currency":"EUR","input_token":0.673,"output_token":3.366,"audio_cost":1.683e-05,"cache_read_cost":0.067,"cache_write_cost":0.034},"providers":["google","google_priority"],"context_size":1048576,"max_output_tokens":65535,"tags":["Instruct","Tools","Image","Reasoning","Audio","Document"],"input_modalities":["text","image","audio","file"],"output_modalities":["text"],"supported_features":["json_mode","reasoning","tools"],"model_series":["gemini-3-flash"],"providers_details":{"google":{"pricing":{"currency":"EUR","input_token":0.673,"output_token":3.366,"audio_cost":1.683e-05,"cache_read_cost":0.067,"cache_write_cost":0.034},"quantization":null,"context_size":1048576,"max_output_tokens":65535,"supported_features":["json_mode","reasoning","tools"],"metrics":{"throughput":221.54,"latency":6.1,"ttft":3.37,"cache_hit_rate":0.52,"uptime":1.0}},"google_priority":{"pricing":{"currency":"EUR","input_token":1.333,"output_token":6.665,"audio_cost":3.332e-05,"cache_read_cost":0.133,"cache_write_cost":0.898},"quantization":null,"context_size":1048576,"max_output_tokens":65535,"supported_features":["json_mode","reasoning","tools"],"metrics":{"throughput":229.6,"latency":5.23,"ttft":4.8,"cache_hit_rate":0.41,"uptime":1.0}}}},{"id":"gemini-3.6-flash","object":"model","created":1786689482,"owned_by":"Google","description":"Gemini 3.6 Flash is a high-efficiency model optimized for multi-step workflows, coding, and improved reasoning performance.","pricing":{"currency":"EUR","input_token":0.673,"output_token":3.366,"audio_cost":1.683e-05,"cache_read_cost":0.067,"cache_write_cost":0.034},"providers":["google","google_priority"],"context_size":1048576,"max_output_tokens":65535,"tags":["Instruct","Tools","Image","Reasoning","Audio"],"input_modalities":["text","image","audio"],"output_modalities":["text"],"supported_features":["json_mode","reasoning","tools"],"model_series":["gemini-3-flash"],"providers_details":{"google":{"pricing":{"currency":"EUR","input_token":0.673,"output_token":3.366,"audio_cost":1.683e-05,"cache_read_cost":0.067,"cache_write_cost":0.034},"quantization":null,"context_size":1048576,"max_output_tokens":65535,"supported_features":["json_mode","reasoning","tools"],"metrics":{"throughput":637.07,"latency":8.54,"ttft":6.84,"cache_hit_rate":0.36,"uptime":1.0}},"google_priority":{"pricing":{"currency":"EUR","input_token":1.333,"output_token":6.665,"audio_cost":3.332e-05,"cache_read_cost":0.133,"cache_write_cost":0.898},"quantization":null,"context_size":1048576,"max_output_tokens":65535,"supported_features":["json_mode","reasoning","tools"],"metrics":{"throughput":1100.33,"latency":7.97,"ttft":5.65,"cache_hit_rate":0.18,"uptime":1.0}}}},{"id":"qwen3.8-2.4t-a95b","object":"model","created":1786685762,"owned_by":"Alibaba Cloud","description":"Qwen3.8-2.4T-A95B is an open-weight flagship MoE model built for advanced reasoning, coding, and long-horizon agentic workflows.","pricing":{"currency":"EUR","input_token":2.244,"output_token":5.386,"cache_read_cost":0.561},"providers":["tensorix"],"context_size":262144,"max_output_tokens":262144,"tags":["Instruct","Reasoning","Tools"],"input_modalities":["text"],"output_modalities":["text"],"supported_features":["json_mode","reasoning","tools"],"model_series":["qwen-3-2.4t"],"providers_details":{"tensorix":{"pricing":{"currency":"EUR","input_token":2.244,"output_token":5.386,"cache_read_cost":0.561},"quantization":"nvfp4","context_size":262144,"max_output_tokens":262144,"supported_features":["json_mode","reasoning","tools"],"metrics":{"throughput":240.68,"latency":26.27,"ttft":1.45,"cache_hit_rate":0.69,"uptime":1.0}}}},{"id":"gemini-3.5-flash-lite","object":"model","created":1786084682,"owned_by":"Google","description":"Gemini 3.5 Flash Lite is a fast, low-cost multimodal model optimized for high-volume, low-latency workloads.","pricing":{"currency":"EUR","input_token":0.296,"output_token":2.468,"audio_cost":6.73e-06,"cache_read_cost":0.03,"cache_write_cost":0.074},"providers":["google","google_priority"],"context_size":1048576,"max_output_tokens":65535,"tags":["Instruct","Tools","Image","Reasoning","Audio","Document"],"input_modalities":["text","image","audio","file"],"output_modalities":["text"],"supported_features":["json_mode","reasoning","tools"],"model_series":["gemini-3-flash-lite"],"providers_details":{"google":{"pricing":{"currency":"EUR","input_token":0.296,"output_token":2.468,"audio_cost":6.73e-06,"cache_read_cost":0.03,"cache_write_cost":0.074},"quantization":null,"context_size":1048576,"max_output_tokens":65535,"supported_features":["json_mode","reasoning","tools"],"metrics":{"throughput":190.22,"latency":2.56,"ttft":0.48,"cache_hit_rate":0.48,"uptime":1.0}},"google_priority":{"pricing":{"currency":"EUR","input_token":0.533,"output_token":4.443,"audio_cost":1.332e-05,"cache_read_cost":0.053,"cache_write_cost":0.153},"quantization":null,"context_size":1048576,"max_output_tokens":65535,"supported_features":["json_mode","reasoning","tools"],"metrics":{"throughput":143.52,"latency":6.14,"ttft":5.72,"cache_hit_rate":0.0,"uptime":1.0}}}},{"id":"deepseek-v4-flash-0731","object":"model","created":1785566282,"owned_by":"Deepseek","description":"DeepSeek V4 Flash 0731 is an efficient MoE model optimized for fast reasoning, coding, and enhanced agentic workflows.","pricing":{"currency":"EUR","input_token":0.081,"output_token":0.153,"cache_read_cost":0.013},"providers":["tensorix","inceptron","scaleway","aki","entrim","sference"],"context_size":1048576,"max_output_tokens":1048576,"tags":["Instruct","Tools","Reasoning"],"input_modalities":["text"],"output_modalities":["text"],"supported_features":["json_mode","reasoning","tools"],"model_series":["deepseek-v4-flash"],"providers_details":{"tensorix":{"pricing":{"currency":"EUR","input_token":0.224,"output_token":0.269,"cache_read_cost":0.056},"quantization":"fp4","context_size":1048576,"max_output_tokens":64000,"supported_features":["json_mode","reasoning","tools"],"metrics":{"throughput":82.74,"latency":9.01,"ttft":2.0,"cache_hit_rate":0.54,"uptime":1.0}},"inceptron":{"pricing":{"currency":"EUR","input_token":0.074,"output_token":0.371,"cache_read_cost":0.054},"quantization":"fp4","context_size":1048576,"max_output_tokens":1048576,"supported_features":["json_mode","reasoning","tools"],"metrics":{"throughput":93.32,"latency":9.46,"ttft":2.01,"cache_hit_rate":0.87,"uptime":0.88}},"scaleway":{"pricing":{"currency":"EUR","input_token":0.4,"output_token":0.8,"cache_read_cost":0.08},"quantization":null,"context_size":256000,"max_output_tokens":32768,"supported_features":["json_mode","reasoning","tools"],"metrics":{"throughput":210.85,"latency":5.62,"ttft":2.61,"cache_hit_rate":0.42,"uptime":1.0}},"aki":{"pricing":{"currency":"EUR","input_token":0.2,"output_token":0.5,"cache_read_cost":0.1},"quantization":"fp16","context_size":1048576,"max_output_tokens":81920,"supported_features":["json_mode","reasoning","tools"],"metrics":{"throughput":60.5,"latency":12.96,"ttft":5.29,"cache_hit_rate":0.4,"uptime":0.98}},"entrim":{"pricing":{"currency":"EUR","input_token":0.081,"output_token":0.153,"cache_read_cost":0.013},"quantization":"fp8","context_size":1048576,"max_output_tokens":393216,"supported_features":["json_mode","reasoning","tools"],"metrics":{"throughput":99.79,"latency":4.28,"ttft":1.47,"cache_hit_rate":0.92,"uptime":0.95}},"sference":{"pricing":{"currency":"EUR","input_token":0.251,"output_token":0.503,"cache_read_cost":0.063},"quantization":"nvfp4","context_size":1048576,"max_output_tokens":1048576,"supported_features":["json_mode","reasoning","tools"],"metrics":{"throughput":342.62,"latency":1.92,"ttft":2.35,"cache_hit_rate":0.77,"uptime":0.99}}}},{"id":"kimi-k3","object":"model","created":1785137882,"owned_by":"Moonshot AI","description":"Kimi K3 is a frontier open-weight multimodal model built for long-horizon reasoning, coding, and advanced agentic workflows.","pricing":{"currency":"EUR","input_token":2.693,"output_token":13.464},"providers":["nebius","tensorix","berget","arkane","sference"],"context_size":1048576,"max_output_tokens":1048576,"tags":["Instruct","Tools","Reasoning","Image"],"input_modalities":["text","image"],"output_modalities":["text"],"supported_features":["json_mode","reasoning","tools"],"model_series":["kimi-k3"],"providers_details":{"nebius":{"pricing":{"currency":"EUR","input_token":2.693,"output_token":13.464},"quantization":"fp4","context_size":1024000,"max_output_tokens":1024000,"supported_features":["json_mode","reasoning","tools"],"metrics":{"throughput":163.16,"latency":13.05,"ttft":1.6,"cache_hit_rate":0.65,"uptime":1.0}},"tensorix":{"pricing":{"currency":"EUR","input_token":2.693,"output_token":13.464,"cache_read_cost":0.673},"quantization":"mxfp4","context_size":1048576,"max_output_tokens":1048576,"supported_features":["json_mode","reasoning","tools"],"metrics":{"throughput":60.26,"latency":32.35,"ttft":2.97,"cache_hit_rate":0.56,"uptime":1.0}},"berget":{"pricing":{"currency":"EUR","input_token":2.693,"output_token":13.464},"quantization":"mxfp4","context_size":1000000,"max_output_tokens":327000,"supported_features":["json_mode","reasoning","tools"],"metrics":{"throughput":158.52,"latency":28.11,"ttft":2.38,"cache_hit_rate":0.6,"uptime":1.0}},"arkane":{"pricing":{"currency":"EUR","input_token":2.693,"output_token":13.464,"cache_read_cost":0.269},"quantization":null,"context_size":1000000,"max_output_tokens":500000,"supported_features":["json_mode","reasoning","tools"],"metrics":{"throughput":347.0,"latency":10.88,"ttft":5.89,"cache_hit_rate":0.92,"uptime":1.0}},"sference":{"pricing":{"currency":"EUR","input_token":2.693,"output_token":13.464,"cache_read_cost":0.404},"quantization":"nvfp4","context_size":1048576,"max_output_tokens":1048576,"supported_features":["json_mode","reasoning","tools"],"metrics":{"throughput":129.8,"latency":15.71,"ttft":1.23,"cache_hit_rate":0.54,"uptime":1.0}}}},{"id":"claude-opus-5","object":"model","created":1785133800,"owned_by":"Anthropic","description":"Claude Opus 5 is Anthropic’s most advanced Opus model, built for long-running agents, complex coding, and high-level professional workflows.","pricing":{"currency":"EUR","input_token":4.937,"output_token":24.684,"cache_read_cost":0.494,"cache_write_cost":6.171},"providers":["amazon_ireland","amazon_paris","google"],"context_size":1000000,"max_output_tokens":128000,"tags":["Instruct","Tools","Image","Reasoning","Document"],"input_modalities":["text","image","file"],"output_modalities":["text"],"supported_features":["json_mode","reasoning","tools"],"model_series":["claude-opus-5"],"providers_details":{"amazon_ireland":{"pricing":{"currency":"EUR","input_token":4.937,"output_token":24.684,"cache_read_cost":0.494,"cache_write_cost":6.171},"quantization":null,"context_size":1000000,"max_output_tokens":128000,"supported_features":["json_mode","reasoning","tools"],"metrics":{"throughput":103.54,"latency":34.2,"ttft":51.12,"cache_hit_rate":0.16,"uptime":0.92}},"amazon_paris":{"pricing":{"currency":"EUR","input_token":4.937,"output_token":24.684,"cache_read_cost":0.494,"cache_write_cost":6.171},"quantization":null,"context_size":1000000,"max_output_tokens":128000,"supported_features":["json_mode","reasoning","tools"],"metrics":{"throughput":204.38,"latency":14.21,"ttft":25.75,"cache_hit_rate":0.71,"uptime":0.98}},"google":{"pricing":{"currency":"EUR","input_token":4.937,"output_token":24.684,"cache_read_cost":0.494,"cache_write_cost":6.171},"quantization":null,"context_size":1000000,"max_output_tokens":128000,"supported_features":["json_mode","reasoning","tools"],"metrics":{"throughput":125.74,"latency":16.62,"ttft":9.82,"cache_hit_rate":0.4,"uptime":0.39}}}},{"id":"gpt-5.6-sol","object":"model","created":1783947600,"owned_by":"OpenAI","description":"GPT-5.6 Sol is the flagship model of the GPT-5.6 family, built for advanced reasoning, deep research, and complex multi-step execution.","pricing":{"currency":"EUR","input_token":3.949,"output_token":19.747,"cache_read_cost":0.395,"cache_write_cost":4.937},"providers":["azure_sc","azure_spc"],"context_size":1050000,"max_output_tokens":128000,"tags":["Instruct","Reasoning","Tools","Image","Document"],"input_modalities":["text","image","file"],"output_modalities":["text"],"supported_features":["json_mode","reasoning","tools"],"model_series":["gpt-5-sol","gpt-5"],"providers_details":{"azure_sc":{"pricing":{"currency":"EUR","input_token":3.949,"output_token":19.747,"cache_read_cost":0.395,"cache_write_cost":4.937},"quantization":null,"context_size":1050000,"max_output_tokens":128000,"supported_features":["json_mode","reasoning","tools"],"metrics":{"throughput":136.74,"latency":6.21,"ttft":3.15,"cache_hit_rate":0.86,"uptime":1.0}},"azure_spc":{"pricing":{"currency":"EUR","input_token":3.949,"output_token":19.747,"cache_read_cost":0.395,"cache_write_cost":4.937},"quantization":null,"context_size":1050000,"max_output_tokens":128000,"supported_features":["json_mode","reasoning","tools"],"metrics":{"throughput":148.08,"latency":9.77,"ttft":1.94,"cache_hit_rate":0.76,"uptime":1.0}}}},{"id":"gpt-5.6-terra","object":"model","created":1783944000,"owned_by":"OpenAI","description":"GPT-5.6 Terra is a balanced mid-tier model optimized for professional work, combining strong reasoning, coding ability, and efficiency.","pricing":{"currency":"EUR","input_token":1.975,"output_token":11.848,"cache_read_cost":0.197,"cache_write_cost":2.468},"providers":["azure_sc","azure_spc"],"context_size":1050000,"max_output_tokens":128000,"tags":["Instruct","Reasoning","Tools","Image","Document"],"input_modalities":["text","image","file"],"output_modalities":["text"],"supported_features":["json_mode","reasoning","tools"],"model_series":["gpt-5-terra","gpt-5"],"providers_details":{"azure_sc":{"pricing":{"currency":"EUR","input_token":1.975,"output_token":11.848,"cache_read_cost":0.197,"cache_write_cost":2.468},"quantization":null,"context_size":1050000,"max_output_tokens":128000,"supported_features":["json_mode","reasoning","tools"],"metrics":{"throughput":138.17,"latency":2.37,"ttft":1.56,"cache_hit_rate":0.66,"uptime":1.0}},"azure_spc":{"pricing":{"currency":"EUR","input_token":1.975,"output_token":11.848,"cache_read_cost":0.197,"cache_write_cost":2.468},"quantization":null,"context_size":1050000,"max_output_tokens":128000,"supported_features":["json_mode","reasoning","tools"],"metrics":{"throughput":76.8,"latency":1.98,"ttft":1.14,"cache_hit_rate":0.25,"uptime":1.0}}}},{"id":"gpt-5.6-luna","object":"model","created":1783940400,"owned_by":"OpenAI","description":"GPT-5.6 Luna is a fast, cost-efficient model optimized for high-volume tasks, real-time interactions, and lightweight reasoning.","pricing":{"currency":"EUR","input_token":0.197,"output_token":1.185,"cache_read_cost":0.02,"cache_write_cost":0.247},"providers":["azure_sc","azure_spc"],"context_size":1050000,"max_output_tokens":128000,"tags":["Instruct","Reasoning","Tools","Image","Document"],"input_modalities":["text","image","file"],"output_modalities":["text"],"supported_features":["json_mode","reasoning","tools"],"model_series":["gpt-5-luna","gpt-5"],"providers_details":{"azure_sc":{"pricing":{"currency":"EUR","input_token":0.197,"output_token":1.185,"cache_read_cost":0.02,"cache_write_cost":0.247},"quantization":null,"context_size":1050000,"max_output_tokens":128000,"supported_features":["json_mode","reasoning","tools"],"metrics":{"throughput":190.28,"latency":9.67,"ttft":1.34,"cache_hit_rate":0.74,"uptime":0.99}},"azure_spc":{"pricing":{"currency":"EUR","input_token":0.197,"output_token":1.185,"cache_read_cost":0.02,"cache_write_cost":0.247},"quantization":null,"context_size":1050000,"max_output_tokens":128000,"supported_features":["json_mode","reasoning","tools"],"metrics":{"throughput":198.4,"latency":9.31,"ttft":1.25,"cache_hit_rate":0.75,"uptime":0.99}}}},{"id":"apertus-70b","object":"model","created":1783507082,"owned_by":"Swiss AI Initiative","description":"Apertus 70B is an open, multilingual language model designed for research, long-context reasoning, and sovereignty-focused AI systems.","pricing":{"currency":"EUR","input_token":1.25,"output_token":2.0},"providers":["aki"],"context_size":65536,"max_output_tokens":16384,"tags":["Instruct","Reasoning","Tools"],"input_modalities":["text"],"output_modalities":["text"],"supported_features":["reasoning","tools"],"model_series":["apertus-70b"],"providers_details":{"aki":{"pricing":{"currency":"EUR","input_token":1.25,"output_token":2.0},"quantization":"fp8","context_size":65536,"max_output_tokens":16384,"supported_features":["reasoning","tools"],"metrics":{"throughput":22.42,"latency":11.81,"ttft":0.58,"cache_hit_rate":0.52,"uptime":1.0}}}},{"id":"claude-sonnet-5","object":"model","created":1782981482,"owned_by":"Anthropic","description":"Claude Sonnet 5 is a high-performance model optimized for coding, agent workflows, and everyday professional tasks with strong efficiency and scalability.","pricing":{"currency":"EUR","input_token":1.975,"output_token":9.874,"cache_read_cost":0.197,"cache_write_cost":2.468},"providers":["google","amazon_ireland","amazon_paris"],"context_size":1000000,"max_output_tokens":128000,"tags":["Instruct","Tools","Image","Reasoning","Document"],"input_modalities":["text","image","file"],"output_modalities":["text"],"supported_features":["json_mode","reasoning","tools"],"model_series":["claude-sonnet-5"],"providers_details":{"google":{"pricing":{"currency":"EUR","input_token":1.975,"output_token":9.874,"cache_read_cost":0.197,"cache_write_cost":2.468},"quantization":null,"context_size":1000000,"max_output_tokens":128000,"supported_features":["json_mode","reasoning","tools"],"metrics":{"throughput":101.77,"latency":10.24,"ttft":4.1,"cache_hit_rate":0.78,"uptime":0.99}},"amazon_ireland":{"pricing":{"currency":"EUR","input_token":1.975,"output_token":9.874,"cache_read_cost":0.197,"cache_write_cost":2.468},"quantization":null,"context_size":1000000,"max_output_tokens":128000,"supported_features":["json_mode","reasoning","tools"],"metrics":{"throughput":249.2,"latency":11.22,"ttft":10.57,"cache_hit_rate":0.83,"uptime":0.93}},"amazon_paris":{"pricing":{"currency":"EUR","input_token":1.975,"output_token":9.874,"cache_read_cost":0.197,"cache_write_cost":2.468},"quantization":null,"context_size":1000000,"max_output_tokens":128000,"supported_features":["json_mode","reasoning","tools"],"metrics":{"throughput":107.48,"latency":9.27,"ttft":11.08,"cache_hit_rate":0.63,"uptime":0.99}}}},{"id":"gemma-4-31b-it","object":"model","created":1781854682,"owned_by":"Google","description":"Gemma 4 31B IT is a multimodal instruction-tuned model from Google DeepMind designed for strong reasoning, coding, and vision-language tasks.","pricing":{"currency":"EUR","input_token":0.2,"output_token":0.35},"providers":["infercom","berget"],"context_size":262000,"max_output_tokens":262000,"tags":["Instruct","Image","Tools","Reasoning"],"input_modalities":["text","image"],"output_modalities":["text"],"supported_features":["json_mode","reasoning","tools"],"model_series":["gemma-4-31b"],"providers_details":{"infercom":{"pricing":{"currency":"EUR","input_token":0.2,"output_token":0.35},"quantization":"bf16","context_size":128000,"max_output_tokens":32768,"supported_features":["json_mode","reasoning","tools"],"metrics":{"throughput":62.95,"latency":3.35,"ttft":1.86,"cache_hit_rate":null,"uptime":1.0}},"berget":{"pricing":{"currency":"EUR","input_token":0.25,"output_token":0.5},"quantization":"fp16","context_size":262000,"max_output_tokens":262000,"supported_features":["json_mode","reasoning","tools"],"metrics":{"throughput":56.34,"latency":4.46,"ttft":1.73,"cache_hit_rate":0.6,"uptime":1.0}}}},{"id":"glm-5.2","object":"model","created":1781686682,"owned_by":"Z.ai","description":"GLM-5.2 is a flagship model built for long-horizon agentic tasks with stable 1M-token context and advanced coding capabilities.","pricing":{"currency":"EUR","input_token":0.808,"output_token":2.908,"cache_read_cost":0.17},"providers":["tensorix","scaleway","inceptron","mistral","arkane"],"context_size":1048576,"max_output_tokens":1000000,"tags":["Instruct","Reasoning","Tools"],"input_modalities":["text"],"output_modalities":["text"],"supported_features":["json_mode","reasoning","tools"],"model_series":["glm-5"],"providers_details":{"tensorix":{"pricing":{"currency":"EUR","input_token":1.331,"output_token":3.993,"cache_read_cost":0.337},"quantization":"fp8","context_size":1048576,"max_output_tokens":131072,"supported_features":["json_mode","reasoning","tools"],"metrics":{"throughput":208.6,"latency":4.25,"ttft":1.55,"cache_hit_rate":0.66,"uptime":1.0}},"scaleway":{"pricing":{"currency":"EUR","input_token":1.8,"output_token":5.5,"cache_read_cost":0.36},"quantization":null,"context_size":256000,"max_output_tokens":16384,"supported_features":["json_mode","reasoning","tools"],"metrics":{"throughput":68.38,"latency":5.89,"ttft":1.01,"cache_hit_rate":null,"uptime":0.68}},"inceptron":{"pricing":{"currency":"EUR","input_token":0.808,"output_token":2.908,"cache_read_cost":0.17},"quantization":"fp4","context_size":1000000,"max_output_tokens":1000000,"supported_features":["json_mode","reasoning","tools"],"metrics":{"throughput":133.01,"latency":2.89,"ttft":2.15,"cache_hit_rate":0.55,"uptime":0.61}},"mistral":{"pricing":{"currency":"EUR","input_token":1.309,"output_token":4.114,"cache_read_cost":0.131},"quantization":null,"context_size":1000000,"max_output_tokens":128000,"supported_features":["json_mode","reasoning","tools"],"metrics":{"throughput":173.37,"latency":2.47,"ttft":1.51,"cache_hit_rate":0.75,"uptime":0.99}},"arkane":{"pricing":{"currency":"EUR","input_token":1.167,"output_token":3.77,"cache_read_cost":0.224},"quantization":null,"context_size":200000,"max_output_tokens":100000,"supported_features":["json_mode","reasoning","tools"],"metrics":{"throughput":114.15,"latency":15.27,"ttft":7.02,"cache_hit_rate":1.0,"uptime":1.0}}}},{"id":"kimi-k2.7-code","object":"model","created":1781599082,"owned_by":"Moonshot AI","description":"Kimi K2.7 Code is an open-source agentic coding model optimized for long-horizon software engineering and efficient multi-step execution.","pricing":{"currency":"EUR","input_token":0.589,"output_token":2.962,"cache_read_cost":0.162},"providers":["tensorix","inceptron"],"context_size":262144,"max_output_tokens":262144,"tags":["Instruct","Tools","Reasoning","Image","Document"],"input_modalities":["text","image","file"],"output_modalities":["text"],"supported_features":["json_mode","reasoning","tools"],"model_series":["kimi-k2"],"providers_details":{"tensorix":{"pricing":{"currency":"EUR","input_token":1.109,"output_token":3.993,"cache_read_cost":0.277},"quantization":"int4","context_size":262144,"max_output_tokens":262144,"supported_features":["json_mode","reasoning","tools"],"metrics":{"throughput":236.87,"latency":3.65,"ttft":0.84,"cache_hit_rate":0.98,"uptime":1.0}},"inceptron":{"pricing":{"currency":"EUR","input_token":0.589,"output_token":2.962,"cache_read_cost":0.162},"quantization":"int4","context_size":256000,"max_output_tokens":256000,"supported_features":["json_mode","reasoning","tools"],"metrics":{"throughput":77.62,"latency":11.76,"ttft":1.37,"cache_hit_rate":0.29,"uptime":0.91}}}},{"id":"minimax-m3","object":"model","created":1781509082,"owned_by":"MiniMax AI","description":"MiniMax M3 is a multimodal frontier model built for coding and agentic workflows, featuring ultra-long context and advanced tool-driven reasoning.","pricing":{"currency":"EUR","input_token":0.269,"output_token":1.077,"cache_read_cost":0.054},"providers":["tensorix","arkane"],"context_size":1048576,"max_output_tokens":1048576,"tags":["Instruct","Tools","Reasoning","Image"],"input_modalities":["text","image"],"output_modalities":["text"],"supported_features":["json_mode","reasoning","tools"],"model_series":["minimax-m3"],"providers_details":{"tensorix":{"pricing":{"currency":"EUR","input_token":0.355,"output_token":1.775,"cache_read_cost":0.089},"quantization":"fp8","context_size":1048576,"max_output_tokens":1048576,"supported_features":["json_mode","reasoning","tools"],"metrics":{"throughput":77.73,"latency":6.58,"ttft":1.09,"cache_hit_rate":0.79,"uptime":1.0}},"arkane":{"pricing":{"currency":"EUR","input_token":0.269,"output_token":1.077,"cache_read_cost":0.054},"quantization":null,"context_size":1000000,"max_output_tokens":500000,"supported_features":["json_mode","reasoning","tools"],"metrics":{"throughput":70.53,"latency":7.34,"ttft":2.64,"cache_hit_rate":0.3,"uptime":0.99}}}},{"id":"qwen3.6-27b","object":"model","created":1780392962,"owned_by":"Alibaba Cloud","description":"Qwen3.6 27B is a dense multimodal model offering strong agentic coding and reasoning performance with easy deployment at a practical scale.","pricing":{"currency":"EUR","input_token":0.4,"output_token":2.7},"providers":["ovh"],"context_size":262000,"max_output_tokens":262000,"tags":["Instruct","Reasoning","Tools","Image"],"input_modalities":["text","image"],"output_modalities":["text"],"supported_features":["json_mode","reasoning","tools"],"model_series":["qwen-3-27b"],"providers_details":{"ovh":{"pricing":{"currency":"EUR","input_token":0.4,"output_token":2.7},"quantization":"fp8","context_size":262000,"max_output_tokens":262000,"supported_features":["json_mode","reasoning","tools"],"metrics":{"throughput":83.32,"latency":12.53,"ttft":0.53,"cache_hit_rate":null,"uptime":1.0}}}},{"id":"minicpm-v-4.5","object":"model","created":1780385882,"owned_by":"OpenBMB","description":"MiniCPM-V 4.5 is a compact, high-performance vision-language model excelling in video understanding, OCR, and multimodal reasoning with efficient deployment.","pricing":{"currency":"EUR","input_token":0.584,"output_token":0.985},"providers":["nebius"],"context_size":32000,"max_output_tokens":32000,"tags":["Instruct"],"input_modalities":["text"],"output_modalities":["text"],"supported_features":["json_mode","tools"],"model_series":["minicpm-v-4"],"providers_details":{"nebius":{"pricing":{"currency":"EUR","input_token":0.584,"output_token":0.985},"quantization":"fp16","context_size":32000,"max_output_tokens":32000,"supported_features":["json_mode","tools"],"metrics":{"throughput":157.0,"latency":2.85,"ttft":0.2,"cache_hit_rate":0.24,"uptime":1.0}}}},{"id":"claude-opus4-8","object":"model","created":1780036200,"owned_by":"Anthropic","description":"Claude Opus 4.8 is a top-tier model built for sustained autonomous work, excelling in coding, agents, and complex enterprise tasks.","pricing":{"currency":"EUR","input_token":4.881,"output_token":24.404,"cache_read_cost":0.488,"cache_write_cost":6.101},"providers":["amazon_ireland","amazon_paris","google"],"context_size":1000000,"max_output_tokens":128000,"tags":["Instruct","Tools","Image","Reasoning","Document"],"input_modalities":["text","image","file"],"output_modalities":["text"],"supported_features":["json_mode","reasoning","tools"],"model_series":["claude-opus-4"],"providers_details":{"amazon_ireland":{"pricing":{"currency":"EUR","input_token":4.881,"output_token":24.404,"cache_read_cost":0.488,"cache_write_cost":6.101},"quantization":null,"context_size":1000000,"max_output_tokens":128000,"supported_features":["json_mode","reasoning","tools"],"metrics":{"throughput":69.42,"latency":11.83,"ttft":2.6,"cache_hit_rate":0.04,"uptime":1.0}},"amazon_paris":{"pricing":{"currency":"EUR","input_token":4.881,"output_token":24.404,"cache_read_cost":0.488,"cache_write_cost":6.101},"quantization":null,"context_size":1000000,"max_output_tokens":128000,"supported_features":["json_mode","reasoning","tools"],"metrics":{"throughput":94.64,"latency":16.56,"ttft":4.23,"cache_hit_rate":0.01,"uptime":0.99}},"google":{"pricing":{"currency":"EUR","input_token":4.881,"output_token":24.404,"cache_read_cost":0.488,"cache_write_cost":6.101},"quantization":null,"context_size":1000000,"max_output_tokens":128000,"supported_features":["json_mode","reasoning","tools"],"metrics":{"throughput":73.87,"latency":8.83,"ttft":2.06,"cache_hit_rate":0.77,"uptime":1.0}}}},{"id":"gemini-3.5-flash","object":"model","created":1779259082,"owned_by":"Google","description":"Gemini 3.5 Flash is a high-speed frontier model delivering near-Pro intelligence for agent workflows, coding, and large-scale tasks.","pricing":{"currency":"EUR","input_token":1.48,"output_token":8.886,"audio_cost":3.366e-05,"cache_read_cost":0.148,"cache_write_cost":0.898},"providers":["google"],"context_size":1048576,"max_output_tokens":65535,"tags":["Instruct","Tools","Image","Reasoning","Audio"],"input_modalities":["text","image","audio"],"output_modalities":["text"],"supported_features":["json_mode","reasoning","tools"],"model_series":["gemini-3-flash"],"providers_details":{"google":{"pricing":{"currency":"EUR","input_token":1.48,"output_token":8.886,"audio_cost":3.366e-05,"cache_read_cost":0.148,"cache_write_cost":0.898},"quantization":null,"context_size":1048576,"max_output_tokens":65535,"supported_features":["json_mode","reasoning","tools"],"metrics":{"throughput":735.18,"latency":9.75,"ttft":6.9,"cache_hit_rate":0.0,"uptime":1.0}}}},{"id":"gemini-3.1-flash-lite","object":"model","created":1779107882,"owned_by":"Google","description":"Gemini 3.1 Flash Lite is a highly efficient multimodal model optimized for ultra-low latency and cost-sensitive, high-volume workloads.","pricing":{"currency":"EUR","input_token":0.244,"output_token":1.464,"audio_cost":1.22e-05,"cache_read_cost":0.022,"cache_write_cost":0.074},"providers":["google","google_priority"],"context_size":1048576,"max_output_tokens":65535,"tags":["Instruct","Tools","Image","Reasoning","Audio","Document"],"input_modalities":["text","image","audio","file"],"output_modalities":["text"],"supported_features":["json_mode","reasoning","tools"],"model_series":["gemini-3-flash-lite"],"providers_details":{"google":{"pricing":{"currency":"EUR","input_token":0.244,"output_token":1.464,"audio_cost":1.22e-05,"cache_read_cost":0.022,"cache_write_cost":0.074},"quantization":null,"context_size":1048576,"max_output_tokens":65535,"supported_features":["json_mode","reasoning","tools"],"metrics":{"throughput":135.82,"latency":1.57,"ttft":1.81,"cache_hit_rate":0.0,"uptime":1.0}},"google_priority":{"pricing":{"currency":"EUR","input_token":0.444,"output_token":2.666,"audio_cost":2.475e-05,"cache_read_cost":0.044,"cache_write_cost":0.153},"quantization":null,"context_size":1048576,"max_output_tokens":65535,"supported_features":["json_mode","reasoning","tools"],"metrics":{"throughput":120.09,"latency":1.46,"ttft":1.31,"cache_hit_rate":0.03,"uptime":1.0}}}},{"id":"qwen3.6-35b-a3b","object":"model","created":1778755082,"owned_by":"Alibaba Cloud","description":"Qwen3.6 35B A3B is a lightweight multimodal MoE model focused on efficient agentic coding, long-context reasoning, and visual understanding.","pricing":{"currency":"EUR","input_token":0.15,"output_token":0.5},"providers":["scaleway","aki"],"context_size":262000,"max_output_tokens":32768,"tags":["Instruct","Reasoning","Tools","Image"],"input_modalities":["text","image"],"output_modalities":["text"],"supported_features":["json_mode","reasoning","tools"],"model_series":["qwen-3-35b"],"providers_details":{"scaleway":{"pricing":{"currency":"EUR","input_token":0.25,"output_token":1.5},"quantization":null,"context_size":262000,"max_output_tokens":32768,"supported_features":["json_mode","reasoning","tools"],"metrics":{"throughput":168.73,"latency":4.66,"ttft":0.55,"cache_hit_rate":null,"uptime":1.0}},"aki":{"pricing":{"currency":"EUR","input_token":0.15,"output_token":0.5},"quantization":"fp8","context_size":256000,"max_output_tokens":32768,"supported_features":["json_mode","reasoning","tools"],"metrics":{"throughput":130.68,"latency":9.78,"ttft":1.26,"cache_hit_rate":0.93,"uptime":0.97}}}},{"id":"gemma-4-26b-a4b-it","object":"model","created":1778153882,"owned_by":"Google","description":"Gemma 4 26B A4B is a multimodal Mixture-of-Experts model optimized for fast, efficient inference with strong reasoning, coding, and agentic capabilities.","pricing":{"currency":"EUR","input_token":0.1,"output_token":0.5},"providers":["scaleway","aki"],"context_size":262000,"max_output_tokens":81920,"tags":["Instruct","Image","Tools","Reasoning"],"input_modalities":["text","image"],"output_modalities":["text"],"supported_features":["json_mode","reasoning","tools"],"model_series":["gemma-4-26b"],"providers_details":{"scaleway":{"pricing":{"currency":"EUR","input_token":0.25,"output_token":0.5},"quantization":null,"context_size":262000,"max_output_tokens":32768,"supported_features":["json_mode","reasoning","tools"],"metrics":{"throughput":155.17,"latency":1.43,"ttft":0.27,"cache_hit_rate":null,"uptime":1.0}},"aki":{"pricing":{"currency":"EUR","input_token":0.1,"output_token":0.5},"quantization":"nvfp4","context_size":256000,"max_output_tokens":81920,"supported_features":["json_mode","reasoning","tools"],"metrics":{"throughput":136.53,"latency":2.27,"ttft":1.39,"cache_hit_rate":0.46,"uptime":0.99}}}},{"id":"deepseek-v4-pro","object":"model","created":1777621082,"owned_by":"Deepseek","description":"DeepSeek-V4 is a Mixture-of-Experts model series featuring advanced attention mechanisms and large-scale training for efficient long-context reasoning and strong performance.","pricing":{"currency":"EUR","input_token":1.526,"output_token":3.052,"cache_read_cost":0.135},"providers":["arkane"],"context_size":1000000,"max_output_tokens":384000,"tags":["Instruct","Tools","Reasoning"],"input_modalities":["text"],"output_modalities":["text"],"supported_features":["json_mode","reasoning","tools"],"model_series":["deepseek-v4-pro"],"providers_details":{"arkane":{"pricing":{"currency":"EUR","input_token":1.526,"output_token":3.052,"cache_read_cost":0.135},"quantization":null,"context_size":1000000,"max_output_tokens":384000,"supported_features":["json_mode","reasoning","tools"],"metrics":{"throughput":65.74,"latency":167.64,"ttft":2.99,"cache_hit_rate":0.76,"uptime":0.99}}}},{"id":"mistral-medium-3.5","object":"model","created":1777549993,"owned_by":"Mistral AI","description":"Mistral Medium 3.5 is a frontier multimodal 128B model combining reasoning, coding, and instruction-following with strong agentic performance and efficient deployment.","pricing":{"currency":"EUR","input_token":1.375,"output_token":7.04,"cache_read_cost":0.138},"providers":["mistral","scaleway"],"context_size":256000,"max_output_tokens":256000,"tags":["Instruct","Reasoning","Tools","Image"],"input_modalities":["text","image"],"output_modalities":["text"],"supported_features":["json_mode","reasoning","tools"],"model_series":["mistral-medium"],"providers_details":{"mistral":{"pricing":{"currency":"EUR","input_token":1.375,"output_token":7.04,"cache_read_cost":0.138},"quantization":null,"context_size":256000,"max_output_tokens":256000,"supported_features":["json_mode","reasoning","tools"],"metrics":{"throughput":113.99,"latency":1.43,"ttft":2.1,"cache_hit_rate":0.05,"uptime":1.0}},"scaleway":{"pricing":{"currency":"EUR","input_token":1.5,"output_token":7.5},"quantization":null,"context_size":256000,"max_output_tokens":16384,"supported_features":["json_mode","reasoning","tools"],"metrics":{"throughput":86.92,"latency":3.54,"ttft":0.63,"cache_hit_rate":null,"uptime":1.0}}}},{"id":"gpt-5.4","object":"model","created":1777030682,"owned_by":"OpenAI","description":"GPT-5.4 is a frontier model designed for fast, reliable performance across professional workflows, coding, data analysis, and agentic tasks.","pricing":{"currency":"EUR","input_token":2.601,"output_token":13.872,"cache_read_cost":0.217},"providers":["azure_sc","azure_spc"],"context_size":1050000,"max_output_tokens":128000,"tags":["Instruct","Tools","Image","Reasoning","Document"],"input_modalities":["text","image","file"],"output_modalities":["text"],"supported_features":["json_mode","reasoning","tools"],"model_series":["gpt-5"],"providers_details":{"azure_sc":{"pricing":{"currency":"EUR","input_token":2.601,"output_token":13.872,"cache_read_cost":0.217},"quantization":null,"context_size":1050000,"max_output_tokens":128000,"supported_features":["json_mode","reasoning","tools"],"metrics":{"throughput":71.24,"latency":2.5,"ttft":1.48,"cache_hit_rate":0.43,"uptime":0.99}},"azure_spc":{"pricing":{"currency":"EUR","input_token":2.601,"output_token":13.872,"cache_read_cost":0.217},"quantization":null,"context_size":1050000,"max_output_tokens":128000,"supported_features":["json_mode","reasoning","tools"],"metrics":{"throughput":100.06,"latency":8.75,"ttft":3.21,"cache_hit_rate":0.12,"uptime":1.0}}}},{"id":"kimi-k2.6","object":"model","created":1776929882,"owned_by":"Moonshot AI","description":"Kimi K2.6 is an advanced multimodal agentic model with strong long-horizon coding, instruction following, and autonomous execution capabilities.","pricing":{"currency":"EUR","input_token":0.426,"output_token":2.666,"cache_read_cost":0.09},"providers":["inceptron"],"context_size":256000,"max_output_tokens":256000,"tags":["Instruct","Tools","Reasoning","Image","Document"],"input_modalities":["text","image","file"],"output_modalities":["text"],"supported_features":["json_mode","reasoning","tools"],"model_series":["kimi-k2"],"providers_details":{"inceptron":{"pricing":{"currency":"EUR","input_token":0.426,"output_token":2.666,"cache_read_cost":0.09},"quantization":"int4","context_size":256000,"max_output_tokens":256000,"supported_features":["json_mode","reasoning","tools"],"metrics":{"throughput":125.59,"latency":75.27,"ttft":2.51,"cache_hit_rate":0.55,"uptime":0.95}}}},{"id":"claude-opus4-7","object":"model","created":1776407400,"owned_by":"Anthropic","description":"Claude Opus 4.7 is Anthropic’s most capable general model, optimized for coding, enterprise workflows, multimodal reasoning, and long-running agentic tasks.","pricing":{"currency":"EUR","input_token":4.881,"output_token":24.404,"cache_read_cost":0.488,"cache_write_cost":6.101},"providers":["amazon_ireland","amazon_paris","google"],"context_size":1000000,"max_output_tokens":128000,"tags":["Instruct","Tools","Image","Reasoning","Document"],"input_modalities":["text","image","file"],"output_modalities":["text"],"supported_features":["json_mode","reasoning","tools"],"model_series":["claude-opus-4"],"providers_details":{"amazon_ireland":{"pricing":{"currency":"EUR","input_token":4.881,"output_token":24.404,"cache_read_cost":0.488,"cache_write_cost":6.101},"quantization":null,"context_size":1000000,"max_output_tokens":128000,"supported_features":["json_mode","reasoning","tools"],"metrics":{"throughput":141.29,"latency":33.86,"ttft":1.47,"cache_hit_rate":0.79,"uptime":0.99}},"amazon_paris":{"pricing":{"currency":"EUR","input_token":4.881,"output_token":24.404,"cache_read_cost":0.488,"cache_write_cost":6.101},"quantization":null,"context_size":1000000,"max_output_tokens":128000,"supported_features":["json_mode","reasoning","tools"],"metrics":{"throughput":129.69,"latency":11.33,"ttft":0.98,"cache_hit_rate":0.86,"uptime":0.97}},"google":{"pricing":{"currency":"EUR","input_token":4.881,"output_token":24.404,"cache_read_cost":0.488,"cache_write_cost":6.101},"quantization":null,"context_size":1000000,"max_output_tokens":128000,"supported_features":["json_mode","reasoning","tools"],"metrics":{"throughput":162.89,"latency":11.36,"ttft":1.54,"cache_hit_rate":0.3,"uptime":1.0}}}},{"id":"glm-5.1","object":"model","created":1775735882,"owned_by":"Z.ai","description":"GLM-5.1 is a next-generation flagship model for agentic engineering, excelling in long-horizon reasoning and advanced coding tasks.","pricing":{"currency":"EUR","input_token":1.242,"output_token":3.903},"providers":["nebius"],"context_size":202752,"max_output_tokens":202752,"tags":["Instruct","Reasoning","Tools"],"input_modalities":["text"],"output_modalities":["text"],"supported_features":["json_mode","reasoning","tools"],"model_series":["glm-5"],"providers_details":{"nebius":{"pricing":{"currency":"EUR","input_token":1.242,"output_token":3.903},"quantization":"fp8","context_size":202752,"max_output_tokens":202752,"supported_features":["json_mode","reasoning","tools"],"metrics":{"throughput":41.2,"latency":26.65,"ttft":1.58,"cache_hit_rate":0.79,"uptime":1.0}}}},{"id":"qwen3.5-9b","object":"model","created":1775727362,"owned_by":"Alibaba Cloud","description":"Qwen3.5-9B is a dense multimodal model delivering strong reasoning, coding, and visual understanding with efficient performance and long context.","pricing":{"currency":"EUR","input_token":0.1,"output_token":0.15},"providers":["ovh","ionos"],"context_size":262000,"max_output_tokens":262000,"tags":["Instruct","Reasoning","Tools","Image"],"input_modalities":["text","image"],"output_modalities":["text"],"supported_features":["json_mode","reasoning","tools"],"model_series":["qwen-3-9b"],"providers_details":{"ovh":{"pricing":{"currency":"EUR","input_token":0.1,"output_token":0.15},"quantization":"bf16","context_size":262000,"max_output_tokens":262000,"supported_features":["json_mode","reasoning","tools"],"metrics":{"throughput":35.1,"latency":59.89,"ttft":0.25,"cache_hit_rate":null,"uptime":1.0}},"ionos":{"pricing":{"currency":"EUR","input_token":0.1,"output_token":0.15},"quantization":"fp8","context_size":262000,"max_output_tokens":262000,"supported_features":["json_mode","reasoning","tools"],"metrics":{"throughput":59.49,"latency":59.84,"ttft":0.47,"cache_hit_rate":null,"uptime":1.0}}}},{"id":"qwen3-coder-next","object":"model","created":1775140682,"owned_by":"Alibaba Cloud","description":"Qwen3 Coder Next 80B is a high-performance coding model optimized for complex software development, agentic workflows, and multi-language programming.","pricing":{"currency":"EUR","input_token":0.15,"output_token":0.8},"providers":["ionos"],"context_size":256000,"max_output_tokens":256000,"tags":["Instruct","Reasoning","Tools"],"input_modalities":["text"],"output_modalities":["text"],"supported_features":["json_mode","reasoning","tools"],"model_series":["qwen-3-coder-next"],"providers_details":{"ionos":{"pricing":{"currency":"EUR","input_token":0.15,"output_token":0.8},"quantization":"int4","context_size":256000,"max_output_tokens":256000,"supported_features":["json_mode","reasoning","tools"],"metrics":{"throughput":106.23,"latency":1.28,"ttft":0.38,"cache_hit_rate":null,"uptime":1.0}}}},{"id":"qwen3.5-397b-a17b","object":"model","created":1773837482,"owned_by":"Alibaba Cloud","description":"Qwen3.5-397B-A17B is a large-scale multimodal model with a hybrid Mixture-of-Experts architecture, delivering state-of-the-art performance across chat, RAG, vision-language, video understanding, and agentic workflows.","pricing":{"currency":"EUR","input_token":0.6,"output_token":3.6},"providers":["scaleway","ovh","ionos"],"context_size":262000,"max_output_tokens":262000,"tags":["Instruct","Reasoning","Tools"],"input_modalities":["text"],"output_modalities":["text"],"supported_features":["json_mode","reasoning","tools"],"model_series":["qwen-3-397b"],"providers_details":{"scaleway":{"pricing":{"currency":"EUR","input_token":0.6,"output_token":3.6},"quantization":null,"context_size":250000,"max_output_tokens":16384,"supported_features":["json_mode","reasoning","tools"],"metrics":{"throughput":84.53,"latency":1.73,"ttft":0.41,"cache_hit_rate":null,"uptime":1.0}},"ovh":{"pricing":{"currency":"EUR","input_token":0.6,"output_token":3.6},"quantization":"fp8","context_size":262000,"max_output_tokens":262000,"supported_features":["json_mode","reasoning","tools"],"metrics":{"throughput":74.98,"latency":1.46,"ttft":0.43,"cache_hit_rate":null,"uptime":1.0}},"ionos":{"pricing":{"currency":"EUR","input_token":0.6,"output_token":3.6},"quantization":"fp8","context_size":262000,"max_output_tokens":262000,"supported_features":["json_mode","reasoning","tools"],"metrics":{"throughput":66.21,"latency":14.33,"ttft":0.81,"cache_hit_rate":null,"uptime":1.0}}}},{"id":"deepseek-v3.2","object":"model","created":1773833882,"owned_by":"Deepseek","description":"DeepSeek V3.2 is a high-efficiency model designed for strong reasoning and agentic tool-use performance, optimized for long-context and scalable post-training.","pricing":{"currency":"EUR","input_token":0.449,"output_token":1.346,"cache_read_cost":0.135},"providers":["arkane"],"context_size":163840,"max_output_tokens":65536,"tags":["Instruct","Tools","Reasoning"],"input_modalities":["text"],"output_modalities":["text"],"supported_features":["json_mode","reasoning","tools"],"model_series":["deepseek-v3"],"providers_details":{"arkane":{"pricing":{"currency":"EUR","input_token":0.449,"output_token":1.346,"cache_read_cost":0.135},"quantization":null,"context_size":163840,"max_output_tokens":65536,"supported_features":["json_mode","reasoning","tools"],"metrics":{"throughput":17.4,"latency":16.12,"ttft":3.43,"cache_hit_rate":0.65,"uptime":0.9}}}},{"id":"mistral-small-2603","object":"model","created":1773741193,"owned_by":"Mistral AI","description":"Mistral Small 4 is a versatile Mixture-of-Experts model that unifies reasoning, multimodal understanding, and agentic coding in a single architecture, optimized for efficiency and long-context interactions.","pricing":{"currency":"EUR","input_token":0.14,"output_token":0.561,"cache_read_cost":0.014},"providers":["mistral","aki"],"context_size":262144,"max_output_tokens":256000,"tags":["Instruct","Tools","Image","Reasoning"],"input_modalities":["text","image"],"output_modalities":["text"],"supported_features":["json_mode","reasoning","tools"],"model_series":["mistral-small"],"providers_details":{"mistral":{"pricing":{"currency":"EUR","input_token":0.14,"output_token":0.561,"cache_read_cost":0.014},"quantization":null,"context_size":256000,"max_output_tokens":256000,"supported_features":["json_mode","reasoning","tools"],"metrics":{"throughput":179.01,"latency":0.75,"ttft":0.43,"cache_hit_rate":0.68,"uptime":1.0}},"aki":{"pricing":{"currency":"EUR","input_token":0.2,"output_token":0.6},"quantization":"fp8","context_size":262144,"max_output_tokens":81920,"supported_features":["json_mode","reasoning","tools"],"metrics":{"throughput":144.5,"latency":2.97,"ttft":0.94,"cache_hit_rate":0.48,"uptime":0.98}}}},{"id":"minimax-m2.5","object":"model","created":1773675482,"owned_by":"MiniMax AI","description":"MiniMax-M2.5 is a state-of-the-art language model optimized for real-world productivity, autonomous agents, coding, and professional office workflows.","pricing":{"currency":"EUR","input_token":0.312,"output_token":1.248},"providers":["amazon_ireland"],"context_size":196000,"max_output_tokens":196000,"tags":["Instruct","Tools","Reasoning"],"input_modalities":["text"],"output_modalities":["text"],"supported_features":["json_mode","reasoning","tools"],"model_series":["minimax-m2"],"providers_details":{"amazon_ireland":{"pricing":{"currency":"EUR","input_token":0.312,"output_token":1.248},"quantization":null,"context_size":196000,"max_output_tokens":196000,"supported_features":["json_mode","reasoning","tools"],"metrics":{"throughput":133.05,"latency":10.8,"ttft":0.44,"cache_hit_rate":null,"uptime":1.0}}}},{"id":"claude-4-6-sonnet","object":"model","created":1771490282,"owned_by":"Anthropic","description":"Claude Sonnet 4.6 delivers frontier-level intelligence at scale, optimized for coding, agents, and enterprise-grade workflows.","pricing":{"currency":"EUR","input_token":2.869,"output_token":14.309,"cache_read_cost":0.287,"cache_write_cost":3.59},"providers":["amazon_ireland","amazon_paris","google"],"context_size":1000000,"max_output_tokens":128000,"tags":["Instruct","Tools","Image","Reasoning","Document"],"input_modalities":["text","image","file"],"output_modalities":["text"],"supported_features":["json_mode","reasoning","tools"],"model_series":["claude-sonnet-4"],"providers_details":{"amazon_ireland":{"pricing":{"currency":"EUR","input_token":2.951,"output_token":14.758,"cache_read_cost":0.293,"cache_write_cost":3.661},"quantization":null,"context_size":1000000,"max_output_tokens":128000,"supported_features":["json_mode","reasoning","tools"],"metrics":{"throughput":143.23,"latency":7.78,"ttft":1.77,"cache_hit_rate":0.57,"uptime":1.0}},"amazon_paris":{"pricing":{"currency":"EUR","input_token":2.951,"output_token":14.758,"cache_read_cost":0.293,"cache_write_cost":3.661},"quantization":null,"context_size":1000000,"max_output_tokens":128000,"supported_features":["json_mode","reasoning","tools"],"metrics":{"throughput":111.48,"latency":8.65,"ttft":1.54,"cache_hit_rate":0.76,"uptime":1.0}},"google":{"pricing":{"currency":"EUR","input_token":2.869,"output_token":14.309,"cache_read_cost":0.287,"cache_write_cost":3.59},"quantization":null,"context_size":1000000,"max_output_tokens":128000,"supported_features":["json_mode","reasoning","tools"],"metrics":{"throughput":67.29,"latency":18.32,"ttft":3.78,"cache_hit_rate":0.39,"uptime":1.0}}}},{"id":"glm-4.7-flash","object":"model","created":1771487882,"owned_by":"Z.ai","description":"GLM-4.7-Flash is a fast, resource-efficient variant of GLM-4.7 optimized for low-latency text and code generation.","pricing":{"currency":"EUR","input_token":0.072,"output_token":0.429},"providers":["amazon_ireland"],"context_size":203000,"max_output_tokens":203000,"tags":["Instruct","Tools"],"input_modalities":["text"],"output_modalities":["text"],"supported_features":["json_mode","tools"],"model_series":["glm-4-flash"],"providers_details":{"amazon_ireland":{"pricing":{"currency":"EUR","input_token":0.072,"output_token":0.429},"quantization":null,"context_size":203000,"max_output_tokens":203000,"supported_features":["json_mode","tools"],"metrics":{"throughput":380.5,"latency":1.19,"ttft":0.65,"cache_hit_rate":null,"uptime":1.0}}}},{"id":"claude-opus4-6","object":"model","created":1770377400,"owned_by":"Anthropic","description":"Claude Opus 4.6 is the next generation of Anthropic’s most intelligent model, setting the global standard for coding, enterprise agents, and professional-grade workflows.","pricing":{"currency":"EUR","input_token":4.769,"output_token":23.843,"cache_read_cost":0.477,"cache_write_cost":5.965},"providers":["amazon_ireland","amazon_paris","google"],"context_size":1000000,"max_output_tokens":128000,"tags":["Instruct","Tools","Image","Reasoning","Document"],"input_modalities":["text","image","file"],"output_modalities":["text"],"supported_features":["json_mode","reasoning","tools"],"model_series":["claude-opus-4"],"providers_details":{"amazon_ireland":{"pricing":{"currency":"EUR","input_token":4.919,"output_token":24.596,"cache_read_cost":0.488,"cache_write_cost":6.101},"quantization":null,"context_size":1000000,"max_output_tokens":128000,"supported_features":["json_mode","reasoning","tools"],"metrics":{"throughput":83.21,"latency":30.0,"ttft":1.78,"cache_hit_rate":0.6,"uptime":1.0}},"amazon_paris":{"pricing":{"currency":"EUR","input_token":4.919,"output_token":24.596,"cache_read_cost":0.488,"cache_write_cost":6.101},"quantization":null,"context_size":1000000,"max_output_tokens":128000,"supported_features":["json_mode","reasoning","tools"],"metrics":{"throughput":99.86,"latency":23.31,"ttft":1.42,"cache_hit_rate":0.4,"uptime":0.9}},"google":{"pricing":{"currency":"EUR","input_token":4.769,"output_token":23.843,"cache_read_cost":0.477,"cache_write_cost":5.965},"quantization":null,"context_size":1000000,"max_output_tokens":128000,"supported_features":["json_mode","reasoning","tools"],"metrics":{"throughput":81.55,"latency":17.53,"ttft":1.65,"cache_hit_rate":0.48,"uptime":1.0}}}},{"id":"minimax-m2","object":"model","created":1770374282,"owned_by":"MiniMax AI","description":"MiniMax-M2 is a large instruction-tuned MoE model optimized for multilingual dialogue, reasoning, and tool-calling, designed as a general-purpose assistant for consumer and enterprise use.","pricing":{"currency":"EUR","input_token":0.313,"output_token":1.261},"providers":["amazon_ireland"],"context_size":400000,"max_output_tokens":196000,"tags":["Instruct","Tools","Reasoning"],"input_modalities":["text"],"output_modalities":["text"],"supported_features":["json_mode","reasoning","tools"],"model_series":["minimax-m2"],"providers_details":{"amazon_ireland":{"pricing":{"currency":"EUR","input_token":0.313,"output_token":1.261},"quantization":null,"context_size":400000,"max_output_tokens":196000,"supported_features":["json_mode","reasoning","tools"],"metrics":{"throughput":142.56,"latency":10.4,"ttft":0.67,"cache_hit_rate":null,"uptime":1.0}}}},{"id":"minimax-m2.1","object":"model","created":1770205082,"owned_by":"MiniMax AI","description":"MiniMax-M2.1 is an open-source agentic coding model designed for polyglot development, precision refactoring, and reliable execution of long, multi-step coding and office workflows.","pricing":{"currency":"EUR","input_token":0.322,"output_token":1.288},"providers":["amazon_ireland"],"context_size":196000,"max_output_tokens":196000,"tags":["Instruct","Tools","Reasoning"],"input_modalities":["text"],"output_modalities":["text"],"supported_features":["json_mode","reasoning","tools"],"model_series":["minimax-m2"],"providers_details":{"amazon_ireland":{"pricing":{"currency":"EUR","input_token":0.322,"output_token":1.288},"quantization":null,"context_size":196000,"max_output_tokens":196000,"supported_features":["json_mode","reasoning","tools"],"metrics":{"throughput":165.98,"latency":8.31,"ttft":0.47,"cache_hit_rate":null,"uptime":1.0}}}},{"id":"qwen3guard-gen-8b","object":"model","created":1770204122,"owned_by":"Alibaba Cloud","description":"Qwen3Guard-Gen-8B is a large-scale multilingual safety moderation model designed for high-accuracy prompt and response classification.","pricing":{"currency":"EUR","input_token":0.0,"output_token":0.0},"providers":["ovh"],"context_size":32000,"max_output_tokens":32000,"tags":["Instruct","Safety-guard"],"input_modalities":["text"],"output_modalities":["text"],"supported_features":[],"model_series":["qwen-3-guard-8b"],"providers_details":{"ovh":{"pricing":{"currency":"EUR","input_token":0.0,"output_token":0.0},"quantization":"fp16","context_size":32000,"max_output_tokens":32000,"supported_features":[],"metrics":{"throughput":29.95,"latency":1.08,"ttft":0.2,"cache_hit_rate":null,"uptime":1.0}}}},{"id":"qwen3guard-gen-0.6b","object":"model","created":1770204002,"owned_by":"Alibaba Cloud","description":"Qwen3Guard-Gen-0.6B is a lightweight multilingual safety moderation model that classifies prompts and responses into safe, controversial, or unsafe categories.","pricing":{"currency":"EUR","input_token":0.0,"output_token":0.0},"providers":["ovh"],"context_size":32000,"max_output_tokens":32000,"tags":["Instruct","Safety-guard"],"input_modalities":["text"],"output_modalities":["text"],"supported_features":[],"model_series":["qwen-3-guard-0.6b"],"providers_details":{"ovh":{"pricing":{"currency":"EUR","input_token":0.0,"output_token":0.0},"quantization":"fp16","context_size":32000,"max_output_tokens":32000,"supported_features":[],"metrics":{"throughput":72.88,"latency":0.16,"ttft":0.11,"cache_hit_rate":null,"uptime":1.0}}}},{"id":"voxtral-small-2507","object":"model","created":1770025993,"owned_by":"Mistral AI","description":"Voxtral Small is a multimodal model with audio input, combining advanced speech capabilities with strong text performance for transcription, translation, and audio understanding.","pricing":{"currency":"EUR","input_token":0.11,"output_token":0.33,"audio_cost":7.37e-05,"cache_read_cost":0.011},"providers":["mistral"],"context_size":32000,"max_output_tokens":32000,"tags":["Instruct","Audio"],"input_modalities":["text","audio"],"output_modalities":["text"],"supported_features":["json_mode","tools"],"model_series":["voxtral-small"],"providers_details":{"mistral":{"pricing":{"currency":"EUR","input_token":0.11,"output_token":0.33,"audio_cost":7.37e-05,"cache_read_cost":0.011},"quantization":null,"context_size":32000,"max_output_tokens":32000,"supported_features":["json_mode","tools"],"metrics":{"throughput":73.3,"latency":0.99,"ttft":0.3,"cache_hit_rate":0.51,"uptime":1.0}}}},{"id":"qwen3-vl-235b-a22b","object":"model","created":1768304282,"owned_by":"Alibaba Cloud","description":"Qwen3 VL 235B A22B is a 235B-parameter MoE vision-language flagship model (≈22B active) designed for frontier-level multimodal understanding across text, images, documents, and long videos.","pricing":{"currency":"EUR","input_token":0.554,"output_token":2.8},"providers":["amazon_ireland"],"context_size":256000,"max_output_tokens":256000,"tags":["Instruct","Reasoning","Tools","Image"],"input_modalities":["text","image"],"output_modalities":["text"],"supported_features":["json_mode","reasoning","tools"],"model_series":["qwen-3-vl-235b"],"providers_details":{"amazon_ireland":{"pricing":{"currency":"EUR","input_token":0.554,"output_token":2.8},"quantization":null,"context_size":256000,"max_output_tokens":256000,"supported_features":["json_mode","reasoning","tools"],"metrics":{"throughput":76.73,"latency":4.11,"ttft":0.48,"cache_hit_rate":null,"uptime":1.0}}}},{"id":"nvidia-nemotron-3-nano-30b-a3b","object":"model","created":1768217882,"owned_by":"Nvidia","description":"Nemotron-Nano-3-30B-A3B is a compact Mixture-of-Experts model optimized for efficient reasoning, chat, and coding, with strong multilingual support and long-context RAG and agent workflows.","pricing":{"currency":"EUR","input_token":0.054,"output_token":0.215},"providers":["nebius","amazon_ireland"],"context_size":256000,"max_output_tokens":256000,"tags":["Instruct","Tools","Reasoning"],"input_modalities":["text"],"output_modalities":["text"],"supported_features":["json_mode","reasoning","tools"],"model_series":["nemotron-3-nano"],"providers_details":{"nebius":{"pricing":{"currency":"EUR","input_token":0.054,"output_token":0.215},"quantization":"fp8","context_size":128000,"max_output_tokens":128000,"supported_features":["json_mode","reasoning","tools"],"metrics":{"throughput":161.78,"latency":4.51,"ttft":0.29,"cache_hit_rate":0.62,"uptime":1.0}},"amazon_ireland":{"pricing":{"currency":"EUR","input_token":0.063,"output_token":0.25},"quantization":null,"context_size":256000,"max_output_tokens":256000,"supported_features":["json_mode","reasoning","tools"],"metrics":{"throughput":216.69,"latency":3.41,"ttft":0.51,"cache_hit_rate":null,"uptime":1.0}}}},{"id":"claude-opus4-5","object":"model","created":1767625200,"owned_by":"Anthropic","description":"Claude Opus 4.5 is Anthropic’s next-generation, most intelligent model, delivering industry-leading performance across coding, agents, computer use, and complex enterprise workflows.","pricing":{"currency":"EUR","input_token":4.769,"output_token":23.849,"cache_read_cost":0.477,"cache_write_cost":5.965},"providers":["amazon_ireland","amazon_paris","google"],"context_size":200000,"max_output_tokens":64000,"tags":["Instruct","Tools","Image","Reasoning","Document"],"input_modalities":["text","image","file"],"output_modalities":["text"],"supported_features":["json_mode","reasoning","tools"],"model_series":["claude-opus-4"],"providers_details":{"amazon_ireland":{"pricing":{"currency":"EUR","input_token":4.919,"output_token":24.596,"cache_read_cost":0.488,"cache_write_cost":6.101},"quantization":null,"context_size":200000,"max_output_tokens":64000,"supported_features":["json_mode","reasoning","tools"],"metrics":{"throughput":94.43,"latency":11.83,"ttft":1.2,"cache_hit_rate":null,"uptime":1.0}},"amazon_paris":{"pricing":{"currency":"EUR","input_token":4.919,"output_token":24.596,"cache_read_cost":0.488,"cache_write_cost":6.101},"quantization":null,"context_size":200000,"max_output_tokens":64000,"supported_features":["json_mode","reasoning","tools"],"metrics":{"throughput":89.73,"latency":9.04,"ttft":1.37,"cache_hit_rate":0.48,"uptime":1.0}},"google":{"pricing":{"currency":"EUR","input_token":4.769,"output_token":23.849,"cache_read_cost":0.477,"cache_write_cost":5.965},"quantization":null,"context_size":200000,"max_output_tokens":64000,"supported_features":["json_mode","reasoning","tools"],"metrics":{"throughput":86.58,"latency":10.42,"ttft":1.02,"cache_hit_rate":0.53,"uptime":1.0}}}},{"id":"devstral-2512","object":"model","created":1765356793,"owned_by":"Mistral AI","description":"Devstral 2 2512 is an enterprise-grade text model designed for advanced coding agents and software engineering workflows.","pricing":{"currency":"EUR","input_token":0.429,"output_token":2.147},"providers":["amazon_ireland"],"context_size":256000,"max_output_tokens":256000,"tags":["Instruct","Tools"],"input_modalities":["text"],"output_modalities":["text"],"supported_features":["json_mode","tools"],"model_series":["devstral"],"providers_details":{"amazon_ireland":{"pricing":{"currency":"EUR","input_token":0.429,"output_token":2.147},"quantization":null,"context_size":256000,"max_output_tokens":256000,"supported_features":["json_mode","tools"],"metrics":{"throughput":108.85,"latency":18.8,"ttft":15.09,"cache_hit_rate":null,"uptime":1.0}}}},{"id":"nova-2-lite","object":"model","created":1764848282,"owned_by":"Amazon","description":"Nova 2 Lite is an advanced multimodal reasoning model that combines efficiency and performance, delivering reliable AI for agentic workflows and enterprise applications.","pricing":{"currency":"EUR","input_token":0.335,"output_token":2.822},"providers":["amazon_ireland","amazon_paris"],"context_size":1000000,"max_output_tokens":65535,"tags":["Instruct","Tools","Image","Reasoning","Document"],"input_modalities":["text","image","file"],"output_modalities":["text"],"supported_features":["json_mode","reasoning","tools"],"model_series":["nova-2-lite"],"providers_details":{"amazon_ireland":{"pricing":{"currency":"EUR","input_token":0.335,"output_token":2.822},"quantization":null,"context_size":1000000,"max_output_tokens":65535,"supported_features":["json_mode","reasoning","tools"],"metrics":{"throughput":219.4,"latency":2.66,"ttft":0.4,"cache_hit_rate":null,"uptime":1.0}},"amazon_paris":{"pricing":{"currency":"EUR","input_token":0.433,"output_token":3.63},"quantization":null,"context_size":1000000,"max_output_tokens":65535,"supported_features":["json_mode","reasoning","tools"],"metrics":{"throughput":259.45,"latency":2.5,"ttft":0.4,"cache_hit_rate":null,"uptime":1.0}}}},{"id":"gpt-oss-safeguard-120b","object":"model","created":1764848282,"owned_by":"OpenAI","description":"GPT OSS Safeguard 120B is a 120B-parameter open-weight safety reasoning model designed for high-accuracy enterprise moderation and policy enforcement.","pricing":{"currency":"EUR","input_token":0.161,"output_token":0.626},"providers":["amazon_ireland"],"context_size":128000,"max_output_tokens":128000,"tags":["Instruct","Reasoning","Tools","Safety-guard"],"input_modalities":["text"],"output_modalities":["text"],"supported_features":["json_mode","reasoning","tools"],"model_series":["gpt-oss-safeguard"],"providers_details":{"amazon_ireland":{"pricing":{"currency":"EUR","input_token":0.161,"output_token":0.626},"quantization":null,"context_size":128000,"max_output_tokens":128000,"supported_features":["json_mode","reasoning","tools"],"metrics":{"throughput":168.95,"latency":1.33,"ttft":0.41,"cache_hit_rate":null,"uptime":1.0}}}},{"id":"mistral-large-2512","object":"model","created":1764762602,"owned_by":"Mistral AI","description":"Mistral Large 3 is a state-of-the-art open-weight multimodal Mixture-of-Experts model with 41B active parameters, delivering frontier-level performance across text and vision tasks.","pricing":{"currency":"EUR","input_token":0.55,"output_token":1.65,"cache_read_cost":0.055},"providers":["mistral"],"context_size":256000,"max_output_tokens":256000,"tags":["Instruct","Tools","Image"],"input_modalities":["text","image"],"output_modalities":["text"],"supported_features":["json_mode","tools"],"model_series":["mistral-large"],"providers_details":{"mistral":{"pricing":{"currency":"EUR","input_token":0.55,"output_token":1.65,"cache_read_cost":0.055},"quantization":null,"context_size":256000,"max_output_tokens":256000,"supported_features":["json_mode","tools"],"metrics":{"throughput":68.55,"latency":4.15,"ttft":1.89,"cache_hit_rate":0.05,"uptime":1.0}}}},{"id":"ministral-8b-2512","object":"model","created":1764762542,"owned_by":"Mistral AI","description":"Ministral 3 8B is a balanced, efficient multimodal model offering strong text and vision capabilities, optimized for edge and local deployment.","pricing":{"currency":"EUR","input_token":0.161,"output_token":0.161},"providers":["mistral","amazon_ireland"],"context_size":256000,"max_output_tokens":256000,"tags":["Instruct","Tools","Image"],"input_modalities":["text","image"],"output_modalities":["text"],"supported_features":["json_mode","tools"],"model_series":["ministral-8b"],"providers_details":{"mistral":{"pricing":{"currency":"EUR","input_token":0.165,"output_token":0.165,"cache_read_cost":0.017},"quantization":null,"context_size":256000,"max_output_tokens":256000,"supported_features":["json_mode","tools"],"metrics":{"throughput":75.54,"latency":1.04,"ttft":0.53,"cache_hit_rate":0.41,"uptime":1.0}},"amazon_ireland":{"pricing":{"currency":"EUR","input_token":0.161,"output_token":0.161},"quantization":null,"context_size":256000,"max_output_tokens":131072,"supported_features":["json_mode","tools"],"metrics":{"throughput":233.61,"latency":3.48,"ttft":0.4,"cache_hit_rate":null,"uptime":1.0}}}},{"id":"ministral-3b-2512","object":"model","created":1764762482,"owned_by":"Mistral AI","description":"Ministral 3 3B is a compact, efficient multimodal model with strong language, vision capabilities, and ideal for custom fine-tuning.","pricing":{"currency":"EUR","input_token":0.11,"output_token":0.11,"cache_read_cost":0.011},"providers":["mistral"],"context_size":256000,"max_output_tokens":256000,"tags":["Instruct","Tools","Image"],"input_modalities":["text","image"],"output_modalities":["text"],"supported_features":["json_mode","tools"],"model_series":["ministral-3b"],"providers_details":{"mistral":{"pricing":{"currency":"EUR","input_token":0.11,"output_token":0.11,"cache_read_cost":0.011},"quantization":null,"context_size":256000,"max_output_tokens":256000,"supported_features":["json_mode","tools"],"metrics":{"throughput":80.91,"latency":0.48,"ttft":0.43,"cache_hit_rate":0.18,"uptime":1.0}}}},{"id":"ministral-14b-2512","object":"model","created":1764761882,"owned_by":"Mistral AI","description":"Ministral 3 14B is a frontier-level 14B multimodal model optimized for local deployment, delivering state-of-the-art text and vision reasoning with a 256K context window and strong agentic capabilities.","pricing":{"currency":"EUR","input_token":0.215,"output_token":0.215},"providers":["mistral","amazon_ireland"],"context_size":256000,"max_output_tokens":256000,"tags":["Instruct","Tools","Image"],"input_modalities":["text","image"],"output_modalities":["text"],"supported_features":["json_mode","tools"],"model_series":["ministral-14b"],"providers_details":{"mistral":{"pricing":{"currency":"EUR","input_token":0.22,"output_token":0.22,"cache_read_cost":0.022},"quantization":null,"context_size":256000,"max_output_tokens":256000,"supported_features":["json_mode","tools"],"metrics":{"throughput":59.8,"latency":3.16,"ttft":0.67,"cache_hit_rate":0.24,"uptime":1.0}},"amazon_ireland":{"pricing":{"currency":"EUR","input_token":0.215,"output_token":0.215},"quantization":null,"context_size":256000,"max_output_tokens":131072,"supported_features":["json_mode","tools"],"metrics":{"throughput":170.9,"latency":5.58,"ttft":0.42,"cache_hit_rate":null,"uptime":1.0}}}},{"id":"gpt-5.1","object":"model","created":1763168400,"owned_by":"OpenAI","description":"GPT-5.1 is an enhanced version of GPT-5, offering clearer responses, adaptive reasoning, improved consistency, expanded context capacity, and better customization for complex and interactive tasks.","pricing":{"currency":"EUR","input_token":1.234,"output_token":9.838,"cache_read_cost":0.14},"providers":["azure_sc"],"context_size":400000,"max_output_tokens":128000,"tags":["Instruct","Reasoning","Tools","Image","Document"],"input_modalities":["text","image","file"],"output_modalities":["text"],"supported_features":["reasoning","tools"],"model_series":["gpt-5"],"providers_details":{"azure_sc":{"pricing":{"currency":"EUR","input_token":1.234,"output_token":9.838,"cache_read_cost":0.14},"quantization":null,"context_size":400000,"max_output_tokens":128000,"supported_features":["reasoning","tools"],"metrics":{"throughput":174.66,"latency":23.59,"ttft":20.09,"cache_hit_rate":0.69,"uptime":0.98}}}},{"id":"nemotron-nano-v2-12b","object":"model","created":1761910682,"owned_by":"Nvidia","description":"NVIDIA Nemotron Nano v2 12B is a 12-billion-parameter multimodal reasoning model designed for advanced video understanding, document intelligence, and visual reasoning, built with a hybrid Transformer-Mamba architecture for high efficiency and low latency.","pricing":{"currency":"EUR","input_token":0.215,"output_token":0.635},"providers":["amazon_ireland"],"context_size":128000,"max_output_tokens":128000,"tags":["Instruct","Tools","Reasoning","Image"],"input_modalities":["text","image"],"output_modalities":["text"],"supported_features":["json_mode","reasoning","tools"],"model_series":["nemotron-2-nano"],"providers_details":{"amazon_ireland":{"pricing":{"currency":"EUR","input_token":0.215,"output_token":0.635},"quantization":"bf16","context_size":128000,"max_output_tokens":128000,"supported_features":["json_mode","reasoning","tools"],"metrics":{"throughput":253.58,"latency":2.02,"ttft":0.6,"cache_hit_rate":null,"uptime":1.0}}}},{"id":"claude-haiku-4-5","object":"model","created":1760547600,"owned_by":"Anthropic","description":"Claude Haiku 4.5 delivers near-frontier performance for coding and agent tasks, optimized for speed and cost to support high-volume and free-tier applications.","pricing":{"currency":"EUR","input_token":0.894,"output_token":4.472,"cache_read_cost":0.089,"cache_write_cost":1.065},"providers":["amazon_ireland","amazon_paris","google"],"context_size":200000,"max_output_tokens":64000,"tags":["Instruct","Tools","Image","Reasoning","Document"],"input_modalities":["text","image","file"],"output_modalities":["text"],"supported_features":["json_mode","reasoning","tools"],"model_series":["claude-haiku-4"],"providers_details":{"amazon_ireland":{"pricing":{"currency":"EUR","input_token":0.894,"output_token":4.472,"cache_read_cost":0.089,"cache_write_cost":1.065},"quantization":null,"context_size":200000,"max_output_tokens":64000,"supported_features":["json_mode","reasoning","tools"],"metrics":{"throughput":204.75,"latency":1.69,"ttft":0.87,"cache_hit_rate":0.6,"uptime":0.98}},"amazon_paris":{"pricing":{"currency":"EUR","input_token":0.894,"output_token":4.472,"cache_read_cost":0.089,"cache_write_cost":1.065},"quantization":null,"context_size":200000,"max_output_tokens":64000,"supported_features":["json_mode","reasoning","tools"],"metrics":{"throughput":203.55,"latency":1.76,"ttft":0.93,"cache_hit_rate":0.56,"uptime":0.99}},"google":{"pricing":{"currency":"EUR","input_token":0.954,"output_token":4.769,"cache_read_cost":0.095,"cache_write_cost":1.196},"quantization":null,"context_size":200000,"max_output_tokens":64000,"supported_features":["json_mode","reasoning","tools"],"metrics":{"throughput":161.74,"latency":1.89,"ttft":0.79,"cache_hit_rate":0.27,"uptime":1.0}}}},{"id":"claude-4-5-sonnet","object":"model","created":1759145882,"owned_by":"Anthropic","description":"Claude Sonnet 4.5 is Anthropic’s most capable model for real-world agents, excelling in coding, computer use, and long-horizon tasks.","pricing":{"currency":"EUR","input_token":2.683,"output_token":13.416,"cache_read_cost":0.293,"cache_write_cost":3.661},"providers":["amazon_ireland","amazon_paris"],"context_size":200000,"max_output_tokens":64000,"tags":["Instruct","Tools","Image","Reasoning","Document"],"input_modalities":["text","image","file"],"output_modalities":["text"],"supported_features":["json_mode","reasoning","tools"],"model_series":["claude-sonnet-4"],"providers_details":{"amazon_ireland":{"pricing":{"currency":"EUR","input_token":2.683,"output_token":13.416,"cache_read_cost":0.293,"cache_write_cost":3.661},"quantization":null,"context_size":200000,"max_output_tokens":64000,"supported_features":["json_mode","reasoning","tools"],"metrics":{"throughput":117.23,"latency":17.59,"ttft":1.24,"cache_hit_rate":0.04,"uptime":1.0}},"amazon_paris":{"pricing":{"currency":"EUR","input_token":2.683,"output_token":13.416,"cache_read_cost":0.293,"cache_write_cost":3.661},"quantization":null,"context_size":200000,"max_output_tokens":64000,"supported_features":["json_mode","reasoning","tools"],"metrics":{"throughput":129.79,"latency":18.01,"ttft":1.35,"cache_hit_rate":0.64,"uptime":1.0}}}},{"id":"gpt-5","object":"model","created":1754528400,"owned_by":"OpenAI","description":"GPT-5 is OpenAI’s most advanced model, built for deep reasoning, high-quality code generation, and complex multi-step tasks.","pricing":{"currency":"EUR","input_token":1.234,"output_token":9.838,"cache_read_cost":0.14},"providers":["azure_sc"],"context_size":400000,"max_output_tokens":128000,"tags":["Instruct","Reasoning","Tools","Image"],"input_modalities":["text","image"],"output_modalities":["text"],"supported_features":["reasoning","tools"],"model_series":["gpt-5"],"providers_details":{"azure_sc":{"pricing":{"currency":"EUR","input_token":1.234,"output_token":9.838,"cache_read_cost":0.14},"quantization":null,"context_size":400000,"max_output_tokens":128000,"supported_features":["reasoning","tools"],"metrics":{"throughput":1158.91,"latency":7.38,"ttft":6.39,"cache_hit_rate":0.49,"uptime":1.0}}}},{"id":"gpt-oss-120b","object":"model","created":1754393882,"owned_by":"OpenAI","description":"GPT Oss 120b is a 120B parameter Mixture-of-Experts model by OpenAI, built for advanced reasoning, agentic tasks, and flexible developer use under Apache 2.0.","pricing":{"currency":"EUR","input_token":0.08,"output_token":0.4},"providers":["scaleway","nebius","ionos","ovh","amazon_ireland","infercom","aki"],"context_size":131000,"max_output_tokens":131000,"tags":["Instruct","Reasoning","Tools"],"input_modalities":["text"],"output_modalities":["text"],"supported_features":["json_mode","reasoning","tools"],"model_series":["gpt-oss-120b"],"providers_details":{"scaleway":{"pricing":{"currency":"EUR","input_token":0.15,"output_token":0.6},"quantization":null,"context_size":128000,"max_output_tokens":8192,"supported_features":["json_mode","reasoning","tools"],"metrics":{"throughput":177.74,"latency":3.18,"ttft":0.6,"cache_hit_rate":null,"uptime":1.0}},"nebius":{"pricing":{"currency":"EUR","input_token":0.134,"output_token":0.537},"quantization":"fp4","context_size":131000,"max_output_tokens":131000,"supported_features":["json_mode","reasoning","tools"],"metrics":{"throughput":408.48,"latency":3.69,"ttft":0.6,"cache_hit_rate":0.34,"uptime":1.0}},"ionos":{"pricing":{"currency":"EUR","input_token":0.15,"output_token":0.65},"quantization":"mxfp4","context_size":128000,"max_output_tokens":128000,"supported_features":["json_mode","reasoning","tools"],"metrics":{"throughput":188.45,"latency":2.79,"ttft":0.43,"cache_hit_rate":null,"uptime":1.0}},"ovh":{"pricing":{"currency":"EUR","input_token":0.08,"output_token":0.4},"quantization":"fp4","context_size":131000,"max_output_tokens":131000,"supported_features":["json_mode","reasoning","tools"],"metrics":{"throughput":67.35,"latency":10.43,"ttft":0.66,"cache_hit_rate":null,"uptime":1.0}},"amazon_ireland":{"pricing":{"currency":"EUR","input_token":0.161,"output_token":0.626},"quantization":null,"context_size":128000,"max_output_tokens":128000,"supported_features":["json_mode","reasoning","tools"],"metrics":{"throughput":210.47,"latency":5.06,"ttft":0.64,"cache_hit_rate":null,"uptime":1.0}},"infercom":{"pricing":{"currency":"EUR","input_token":0.22,"output_token":0.59},"quantization":"mxfp4","context_size":128000,"max_output_tokens":128000,"supported_features":["json_mode","reasoning","tools"],"metrics":{"throughput":575.01,"latency":2.6,"ttft":0.96,"cache_hit_rate":null,"uptime":1.0}},"aki":{"pricing":{"currency":"EUR","input_token":0.15,"output_token":0.55},"quantization":"mxfp4","context_size":128000,"max_output_tokens":32768,"supported_features":["json_mode","reasoning","tools"],"metrics":{"throughput":195.52,"latency":1.99,"ttft":2.5,"cache_hit_rate":0.15,"uptime":1.0}}}},{"id":"qwen3-30b-a3b-instruct-2507","object":"model","created":1753702682,"owned_by":"Alibaba Cloud","description":"Qwen3-30B-A3B-Instruct-2507 is an advanced Mixture-of-Experts model optimized for reasoning, coding, and multilingual instruction following.","pricing":{"currency":"EUR","input_token":0.089,"output_token":0.268},"providers":["nebius"],"context_size":262000,"max_output_tokens":262000,"tags":["Instruct","Tools","Reasoning"],"input_modalities":["text"],"output_modalities":["text"],"supported_features":["json_mode","reasoning","tools"],"model_series":["qwen-3-30b"],"providers_details":{"nebius":{"pricing":{"currency":"EUR","input_token":0.089,"output_token":0.268},"quantization":"fp8","context_size":262000,"max_output_tokens":262000,"supported_features":["json_mode","reasoning","tools"],"metrics":{"throughput":48.37,"latency":3.67,"ttft":0.35,"cache_hit_rate":0.54,"uptime":1.0}}}},{"id":"gpt-oss-20b","object":"model","created":1752838682,"owned_by":"OpenAI","description":"GPT Oss 20B is an open-weight Mixture-of-Experts model from OpenAI, optimized for low-latency inference and local deployment.","pricing":{"currency":"EUR","input_token":0.04,"output_token":0.15},"providers":["ovh","amazon_ireland"],"context_size":131000,"max_output_tokens":131000,"tags":["Instruct","Reasoning","Tools"],"input_modalities":["text"],"output_modalities":["text"],"supported_features":["json_mode","reasoning","tools"],"model_series":["gpt-oss-20b"],"providers_details":{"ovh":{"pricing":{"currency":"EUR","input_token":0.04,"output_token":0.15},"quantization":"fp4","context_size":131000,"max_output_tokens":131000,"supported_features":["json_mode","reasoning","tools"],"metrics":{"throughput":76.47,"latency":20.3,"ttft":0.97,"cache_hit_rate":null,"uptime":1.0}},"amazon_ireland":{"pricing":{"currency":"EUR","input_token":0.072,"output_token":0.313},"quantization":null,"context_size":128000,"max_output_tokens":128000,"supported_features":["json_mode","reasoning","tools"],"metrics":{"throughput":238.82,"latency":2.35,"ttft":0.79,"cache_hit_rate":null,"uptime":1.0}}}},{"id":"mistral-7b-instruct-v0.3","object":"model","created":1748252162,"owned_by":"Mistral AI","description":"Mistral-7B-Instruct-v0.3 model is a fine-tuned version of the Mistral 7B base model, optimized for instruction-following tasks. Released in 2023, it is intended for demonstration purposes and does not include built-in guardrails or moderation features.","pricing":{"currency":"EUR","input_token":0.1,"output_token":0.1},"providers":["ovh"],"context_size":127000,"max_output_tokens":127000,"tags":["Instruct","Tools"],"input_modalities":["text"],"output_modalities":["text"],"supported_features":["json_mode","tools"],"model_series":["mistral-7b"],"providers_details":{"ovh":{"pricing":{"currency":"EUR","input_token":0.1,"output_token":0.1},"quantization":"fp16","context_size":127000,"max_output_tokens":127000,"supported_features":["json_mode","tools"],"metrics":{"throughput":38.3,"latency":7.64,"ttft":0.22,"cache_hit_rate":null,"uptime":1.0}}}},{"id":"mistral-7b-instruct-v0.2","object":"model","created":1748252162,"owned_by":"Mistral AI","description":"Mistral 7B Instruct is a compact, 7B parameter model optimized for fast and efficient text and code generation with a 32K token context window.","pricing":{"currency":"EUR","input_token":0.143,"output_token":0.197},"providers":["amazon_ireland","amazon_paris"],"context_size":32000,"max_output_tokens":8192,"tags":["Instruct","Document"],"input_modalities":["text","file"],"output_modalities":["text"],"supported_features":[],"model_series":["mistral-7b"],"providers_details":{"amazon_ireland":{"pricing":{"currency":"EUR","input_token":0.143,"output_token":0.197},"quantization":null,"context_size":32000,"max_output_tokens":8192,"supported_features":[],"metrics":{"throughput":131.82,"latency":2.3,"ttft":0.2,"cache_hit_rate":null,"uptime":1.0}},"amazon_paris":{"pricing":{"currency":"EUR","input_token":0.179,"output_token":0.233},"quantization":null,"context_size":32000,"max_output_tokens":8192,"supported_features":[],"metrics":{"throughput":134.95,"latency":2.43,"ttft":0.19,"cache_hit_rate":null,"uptime":1.0}}}},{"id":"mistral-small-3.2-24b-instruct-2506","object":"model","created":1748252162,"owned_by":"Mistral AI","description":"Mistral-Small-3.2-24B-Instruct-2506 is a 24B parameter instruction-tuned model with enhanced long-context support (128k) and state-of-the-art vision understanding.","pricing":{"currency":"EUR","input_token":0.09,"output_token":0.28},"providers":["ovh","scaleway","berget"],"context_size":131000,"max_output_tokens":131000,"tags":["Instruct","Tools","Image"],"input_modalities":["text","image"],"output_modalities":["text"],"supported_features":["json_mode","tools"],"model_series":["mistral-small"],"providers_details":{"ovh":{"pricing":{"currency":"EUR","input_token":0.09,"output_token":0.28},"quantization":"bf16","context_size":128000,"max_output_tokens":128000,"supported_features":["json_mode","tools"],"metrics":{"throughput":41.19,"latency":5.78,"ttft":0.36,"cache_hit_rate":null,"uptime":1.0}},"scaleway":{"pricing":{"currency":"EUR","input_token":0.15,"output_token":0.35},"quantization":null,"context_size":128000,"max_output_tokens":8192,"supported_features":["json_mode","tools"],"metrics":{"throughput":127.18,"latency":3.19,"ttft":0.27,"cache_hit_rate":null,"uptime":1.0}},"berget":{"pricing":{"currency":"EUR","input_token":0.3,"output_token":0.3},"quantization":"bf16","context_size":131000,"max_output_tokens":131000,"supported_features":["json_mode","tools"],"metrics":{"throughput":66.98,"latency":1.9,"ttft":0.28,"cache_hit_rate":null,"uptime":1.0}}}},{"id":"mistral-large-2402","object":"model","created":1748252162,"owned_by":"Mistral AI","description":"Mistral Large (24.02) is Mistral AI’s most advanced language model, built for complex multilingual reasoning, code generation, and deep text understanding.","pricing":{"currency":"EUR","input_token":3.846,"output_token":11.627},"providers":["amazon_ireland","amazon_paris"],"context_size":32000,"max_output_tokens":8192,"tags":["Instruct","Tools","Reasoning","Document"],"input_modalities":["text","file"],"output_modalities":["text"],"supported_features":["json_mode","reasoning","tools"],"model_series":["mistral-large"],"providers_details":{"amazon_ireland":{"pricing":{"currency":"EUR","input_token":3.846,"output_token":11.627},"quantization":null,"context_size":32000,"max_output_tokens":8192,"supported_features":["json_mode","reasoning","tools"],"metrics":{"throughput":52.03,"latency":7.03,"ttft":0.22,"cache_hit_rate":null,"uptime":1.0}},"amazon_paris":{"pricing":{"currency":"EUR","input_token":4.648,"output_token":13.951},"quantization":null,"context_size":32000,"max_output_tokens":8192,"supported_features":["json_mode","reasoning","tools"],"metrics":{"throughput":51.71,"latency":7.77,"ttft":0.23,"cache_hit_rate":null,"uptime":1.0}}}},{"id":"pixtral-large-2502","object":"model","created":1748252162,"owned_by":"Mistral AI","description":"Pixtral Large (25.02) is a 124B open-weight multimodal model built on Mistral Large 2, offering advanced image understanding and strong performance across text and code tasks.","pricing":{"currency":"EUR","input_token":1.789,"output_token":5.366},"providers":["amazon_ireland","amazon_paris"],"context_size":128000,"max_output_tokens":128000,"tags":["Instruct","Image","Tools","Reasoning","Document"],"input_modalities":["text","image","file"],"output_modalities":["text"],"supported_features":["json_mode","reasoning","tools"],"model_series":["pixtral-large"],"providers_details":{"amazon_ireland":{"pricing":{"currency":"EUR","input_token":1.789,"output_token":5.366},"quantization":null,"context_size":128000,"max_output_tokens":128000,"supported_features":["json_mode","reasoning","tools"],"metrics":{"throughput":85.14,"latency":5.69,"ttft":0.37,"cache_hit_rate":null,"uptime":1.0}},"amazon_paris":{"pricing":{"currency":"EUR","input_token":1.789,"output_token":5.366},"quantization":null,"context_size":128000,"max_output_tokens":128000,"supported_features":["json_mode","reasoning","tools"],"metrics":{"throughput":93.26,"latency":5.45,"ttft":0.38,"cache_hit_rate":null,"uptime":1.0}}}},{"id":"qwen3-235b-a22b-instruct-2507","object":"model","created":1745840282,"owned_by":"Alibaba Cloud","description":"Qwen3-235B-A22B-Instruct-2507 is a large non-thinking mode model with major improvements in reasoning, instruction following, knowledge coverage, and long-context understanding up to 256K tokens.","pricing":{"currency":"EUR","input_token":0.179,"output_token":0.537},"providers":["scaleway","nebius"],"context_size":262000,"max_output_tokens":262000,"tags":["Instruct","Tools","Reasoning"],"input_modalities":["text"],"output_modalities":["text"],"supported_features":["json_mode","reasoning","tools"],"model_series":["qwen-3-235b"],"providers_details":{"scaleway":{"pricing":{"currency":"EUR","input_token":0.75,"output_token":2.25},"quantization":null,"context_size":250000,"max_output_tokens":8192,"supported_features":["json_mode","reasoning","tools"],"metrics":{"throughput":88.64,"latency":2.49,"ttft":0.16,"cache_hit_rate":null,"uptime":1.0}},"nebius":{"pricing":{"currency":"EUR","input_token":0.179,"output_token":0.537},"quantization":"fp4","context_size":262000,"max_output_tokens":262000,"supported_features":["json_mode","reasoning","tools"],"metrics":{"throughput":57.87,"latency":46.0,"ttft":12.98,"cache_hit_rate":0.36,"uptime":0.98}}}},{"id":"qwen3-coder-30b-a3b-instruct","object":"model","created":1745840282,"owned_by":"Alibaba Cloud","description":"Qwen3 Coder 30b a3b Instruct is a 30.5B parameter open-source Mixture-of-Experts model specialized for coding, tool use, and complex multi-step programming workflows.","pricing":{"currency":"EUR","input_token":0.161,"output_token":0.626},"providers":["scaleway","amazon_ireland"],"context_size":262144,"max_output_tokens":262144,"tags":["Instruct","Tools","Reasoning"],"input_modalities":["text"],"output_modalities":["text"],"supported_features":["json_mode","reasoning","tools"],"model_series":["qwen-3-coder-30b"],"providers_details":{"scaleway":{"pricing":{"currency":"EUR","input_token":0.2,"output_token":0.8},"quantization":null,"context_size":128000,"max_output_tokens":32000,"supported_features":["json_mode","reasoning","tools"],"metrics":{"throughput":159.06,"latency":1.64,"ttft":0.15,"cache_hit_rate":null,"uptime":1.0}},"amazon_ireland":{"pricing":{"currency":"EUR","input_token":0.161,"output_token":0.626},"quantization":null,"context_size":262144,"max_output_tokens":262144,"supported_features":["json_mode","reasoning","tools"],"metrics":{"throughput":296.49,"latency":1.45,"ttft":0.59,"cache_hit_rate":null,"uptime":1.0}}}},{"id":"qwen3-32b","object":"model","created":1745840282,"owned_by":"Alibaba Cloud","description":"Qwen3 delivers cutting-edge advancements in reasoning, agent capabilities, and multilingual support, with seamless mode switching for optimized performance across tasks.","pricing":{"currency":"EUR","input_token":0.161,"output_token":0.626},"providers":["amazon_ireland"],"context_size":16384,"max_output_tokens":16384,"tags":["Instruct","Reasoning","Tools"],"input_modalities":["text"],"output_modalities":["text"],"supported_features":["json_mode","reasoning","tools"],"model_series":["qwen-3-32b"],"providers_details":{"amazon_ireland":{"pricing":{"currency":"EUR","input_token":0.161,"output_token":0.626},"quantization":null,"context_size":16384,"max_output_tokens":16384,"supported_features":["json_mode","reasoning","tools"],"metrics":{"throughput":142.16,"latency":3.21,"ttft":0.4,"cache_hit_rate":null,"uptime":1.0}}}},{"id":"nova-lite-v1","object":"model","created":1744630682,"owned_by":"Amazon","description":"Nova Lite is a fast, low-cost multimodal foundation model capable of reasoning over text, images, and video in 200+ languages.","pricing":{"currency":"EUR","input_token":0.062,"output_token":0.247},"providers":["amazon_ireland","amazon_paris"],"context_size":300000,"max_output_tokens":10000,"tags":["Instruct","Tools","Image","Reasoning","Document"],"input_modalities":["text","image","file"],"output_modalities":["text"],"supported_features":["json_mode","reasoning","tools"],"model_series":["nova-1-lite"],"providers_details":{"amazon_ireland":{"pricing":{"currency":"EUR","input_token":0.062,"output_token":0.247},"quantization":null,"context_size":300000,"max_output_tokens":10000,"supported_features":["json_mode","reasoning","tools"],"metrics":{"throughput":251.05,"latency":2.17,"ttft":0.35,"cache_hit_rate":null,"uptime":1.0}},"amazon_paris":{"pricing":{"currency":"EUR","input_token":0.079,"output_token":0.315},"quantization":null,"context_size":300000,"max_output_tokens":10000,"supported_features":["json_mode","reasoning","tools"],"metrics":{"throughput":235.7,"latency":2.0,"ttft":0.37,"cache_hit_rate":null,"uptime":1.0}}}},{"id":"claude-sonnet-4","object":"model","created":1744630682,"owned_by":"Anthropic","description":"Claude Sonnet 4 is a versatile, mid-sized model optimized for coding, real-time AI assistants, and large-scale content tasks with strong cost-performance balance.","pricing":{"currency":"EUR","input_token":2.683,"output_token":13.416,"cache_read_cost":0.266,"cache_write_cost":3.328},"providers":["amazon_ireland","amazon_paris"],"context_size":200000,"max_output_tokens":65000,"tags":["Instruct","Tools","Image","Reasoning","Document"],"input_modalities":["text","image","file"],"output_modalities":["text"],"supported_features":["json_mode","reasoning","tools"],"model_series":["claude-sonnet-4"],"providers_details":{"amazon_ireland":{"pricing":{"currency":"EUR","input_token":2.683,"output_token":13.416,"cache_read_cost":0.266,"cache_write_cost":3.328},"quantization":null,"context_size":200000,"max_output_tokens":65000,"supported_features":["json_mode","reasoning","tools"],"metrics":{"throughput":100.46,"latency":9.66,"ttft":1.53,"cache_hit_rate":null,"uptime":1.0}},"amazon_paris":{"pricing":{"currency":"EUR","input_token":2.683,"output_token":13.416,"cache_read_cost":0.266,"cache_write_cost":3.328},"quantization":null,"context_size":200000,"max_output_tokens":65000,"supported_features":["json_mode","reasoning","tools"],"metrics":{"throughput":126.84,"latency":10.37,"ttft":2.8,"cache_hit_rate":null,"uptime":1.0}}}},{"id":"gpt-4.1-mini","object":"model","created":1744630682,"owned_by":"OpenAI","description":"gpt-4.1-mini is a cost-effective, low-latency model in the GPT-4.1 series, optimized for coding, instruction following, and multimodal tasks with large-context support.","pricing":{"currency":"EUR","input_token":0.39,"output_token":1.53,"cache_read_cost":0.12},"providers":["azure_sc","azure_spc"],"context_size":1047576,"max_output_tokens":32768,"tags":["Instruct","Tools","Image","Reasoning","Document"],"input_modalities":["text","image","file"],"output_modalities":["text"],"supported_features":["reasoning","tools"],"model_series":["gpt-4-mini"],"providers_details":{"azure_sc":{"pricing":{"currency":"EUR","input_token":0.39,"output_token":1.53,"cache_read_cost":0.12},"quantization":null,"context_size":1047576,"max_output_tokens":32768,"supported_features":["reasoning","tools"],"metrics":{"throughput":90.45,"latency":1.49,"ttft":0.52,"cache_hit_rate":0.01,"uptime":1.0}},"azure_spc":{"pricing":{"currency":"EUR","input_token":0.39,"output_token":1.53,"cache_read_cost":0.12},"quantization":null,"context_size":1047576,"max_output_tokens":32768,"supported_features":["reasoning","tools"],"metrics":{"throughput":60.09,"latency":1.47,"ttft":0.56,"cache_hit_rate":0.05,"uptime":1.0}}}},{"id":"gpt-4.1-nano","object":"model","created":1744630682,"owned_by":"OpenAI","description":"gpt-4.1-nano is an ultra-efficient model in the GPT-4.1 series, offering powerful coding and instruction capabilities with support for text and vision at the lowest cost and latency.","pricing":{"currency":"EUR","input_token":0.1,"output_token":0.39,"cache_read_cost":0.05},"providers":["azure_sc","azure_spc"],"context_size":1047576,"max_output_tokens":32768,"tags":["Instruct","Tools","Image","Reasoning","Document"],"input_modalities":["text","image","file"],"output_modalities":["text"],"supported_features":["reasoning","tools"],"model_series":["gpt-4-nano"],"providers_details":{"azure_sc":{"pricing":{"currency":"EUR","input_token":0.1,"output_token":0.39,"cache_read_cost":0.05},"quantization":null,"context_size":1047576,"max_output_tokens":32768,"supported_features":["reasoning","tools"],"metrics":{"throughput":233.73,"latency":1.07,"ttft":0.42,"cache_hit_rate":0.09,"uptime":1.0}},"azure_spc":{"pricing":{"currency":"EUR","input_token":0.1,"output_token":0.39,"cache_read_cost":0.05},"quantization":null,"context_size":1047576,"max_output_tokens":32768,"supported_features":["reasoning","tools"],"metrics":{"throughput":241.35,"latency":1.32,"ttft":0.54,"cache_hit_rate":0.3,"uptime":1.0}}}},{"id":"nova-micro-v1","object":"model","created":1744630682,"owned_by":"Amazon","description":"Nova Micro is a multilingual text-to-text foundation model with strong reasoning capabilities and broad language coverage across 200+ languages.","pricing":{"currency":"EUR","input_token":0.036,"output_token":0.143},"providers":["amazon_ireland","amazon_paris"],"context_size":128000,"max_output_tokens":10000,"tags":["Instruct","Tools","Image","Reasoning"],"input_modalities":["text","image"],"output_modalities":["text"],"supported_features":["json_mode","reasoning","tools"],"model_series":["nova-1-micro"],"providers_details":{"amazon_ireland":{"pricing":{"currency":"EUR","input_token":0.036,"output_token":0.143},"quantization":null,"context_size":128000,"max_output_tokens":10000,"supported_features":["json_mode","reasoning","tools"],"metrics":{"throughput":214.96,"latency":1.25,"ttft":0.48,"cache_hit_rate":null,"uptime":1.0}},"amazon_paris":{"pricing":{"currency":"EUR","input_token":0.047,"output_token":0.185},"quantization":null,"context_size":128000,"max_output_tokens":10000,"supported_features":["json_mode","reasoning","tools"],"metrics":{"throughput":346.46,"latency":1.44,"ttft":0.59,"cache_hit_rate":null,"uptime":1.0}}}},{"id":"gpt-4.1","object":"model","created":1744630682,"owned_by":"OpenAI","description":"GPT-4.1 is the latest evolution of the GPT-4o model family, offering superior coding, instruction following, and support for up to 1 million input tokens with enhanced multimodal capabilities.","pricing":{"currency":"EUR","input_token":1.968,"output_token":7.872,"cache_read_cost":0.49},"providers":["azure_sc","azure_spc"],"context_size":1047576,"max_output_tokens":32768,"tags":["Instruct","Tools","Image","Reasoning","Document"],"input_modalities":["text","image","file"],"output_modalities":["text"],"supported_features":["reasoning","tools"],"model_series":["gpt-4"],"providers_details":{"azure_sc":{"pricing":{"currency":"EUR","input_token":1.968,"output_token":7.872,"cache_read_cost":0.49},"quantization":null,"context_size":1047576,"max_output_tokens":32768,"supported_features":["reasoning","tools"],"metrics":{"throughput":121.72,"latency":2.1,"ttft":0.47,"cache_hit_rate":0.35,"uptime":1.0}},"azure_spc":{"pricing":{"currency":"EUR","input_token":1.968,"output_token":7.872,"cache_read_cost":0.49},"quantization":null,"context_size":1047576,"max_output_tokens":32768,"supported_features":["reasoning","tools"],"metrics":{"throughput":115.1,"latency":2.1,"ttft":0.54,"cache_hit_rate":0.37,"uptime":1.0}}}},{"id":"nova-pro-v1","object":"model","created":1744630682,"owned_by":"Amazon","description":"Nova Pro is a powerful, multilingual multimodal foundation model that excels at reasoning over text, images, and video across 200+ languages.","pricing":{"currency":"EUR","input_token":0.824,"output_token":3.295},"providers":["amazon_ireland","amazon_paris"],"context_size":300000,"max_output_tokens":10000,"tags":["Instruct","Tools","Image","Reasoning","Document"],"input_modalities":["text","image","file"],"output_modalities":["text"],"supported_features":["json_mode","reasoning","tools"],"model_series":["nova-1-pro"],"providers_details":{"amazon_ireland":{"pricing":{"currency":"EUR","input_token":0.824,"output_token":3.295},"quantization":null,"context_size":300000,"max_output_tokens":10000,"supported_features":["json_mode","reasoning","tools"],"metrics":{"throughput":164.77,"latency":2.57,"ttft":0.4,"cache_hit_rate":null,"uptime":1.0}},"amazon_paris":{"pricing":{"currency":"EUR","input_token":1.055,"output_token":4.221},"quantization":null,"context_size":300000,"max_output_tokens":10000,"supported_features":["json_mode","reasoning","tools"],"metrics":{"throughput":213.13,"latency":2.04,"ttft":0.39,"cache_hit_rate":null,"uptime":1.0}}}},{"id":"mistral-small-2503","object":"model","created":1742482730,"owned_by":"Mistral AI","description":"Combines advanced text and vision capabilities with 24 billion parameters, supporting multilingual tasks and long contexts up to 131k tokens, making it versatile for various applications without sacrificing performance.","pricing":{"currency":"EUR","input_token":0.1,"output_token":0.3},"providers":["ionos"],"context_size":128000,"max_output_tokens":128000,"tags":["Instruct","Image","Tools"],"input_modalities":["text","image"],"output_modalities":["text"],"supported_features":["json_mode","tools"],"model_series":["mistral-small"],"providers_details":{"ionos":{"pricing":{"currency":"EUR","input_token":0.1,"output_token":0.3},"quantization":"fp8","context_size":128000,"max_output_tokens":128000,"supported_features":["json_mode","tools"],"metrics":{"throughput":32.13,"latency":4.52,"ttft":0.37,"cache_hit_rate":null,"uptime":0.9}}}},{"id":"gemini-2.5-flash","object":"model","created":1742125082,"owned_by":"Google","description":"Gemini 2.5 models are advanced thinking models that reason before responding, delivering improved accuracy and performance.","pricing":{"currency":"EUR","input_token":0.268,"output_token":2.236,"audio_cost":2.236e-05,"cache_read_cost":0.026,"cache_write_cost":0.087},"providers":["google"],"context_size":1048576,"max_output_tokens":65535,"tags":["Instruct","Tools","Image","Reasoning","Audio","Document"],"input_modalities":["text","image","audio","file"],"output_modalities":["text"],"supported_features":["json_mode","reasoning","tools"],"model_series":["gemini-2-flash"],"providers_details":{"google":{"pricing":{"currency":"EUR","input_token":0.268,"output_token":2.236,"audio_cost":2.236e-05,"cache_read_cost":0.026,"cache_write_cost":0.087},"quantization":null,"context_size":1048576,"max_output_tokens":65535,"supported_features":["json_mode","reasoning","tools"],"metrics":{"throughput":171.24,"latency":9.68,"ttft":25.62,"cache_hit_rate":0.09,"uptime":1.0}}}},{"id":"gemini-2.5-pro","object":"model","created":1742125082,"owned_by":"Google","description":"Gemini 2.5 Pro is the most advanced reasoning model in the Gemini series, excelling at complex problem-solving across multiple data types.","pricing":{"currency":"EUR","input_token":1.342,"output_token":8.944,"audio_cost":3.4e-05,"cache_read_cost":0.217,"cache_write_cost":0.39},"providers":["google"],"context_size":1048576,"max_output_tokens":65535,"tags":["Instruct","Tools","Image","Reasoning","Audio"],"input_modalities":["text","image","audio"],"output_modalities":["text"],"supported_features":["json_mode","reasoning","tools"],"model_series":["gemini-2-pro"],"providers_details":{"google":{"pricing":{"currency":"EUR","input_token":1.342,"output_token":8.944,"audio_cost":3.4e-05,"cache_read_cost":0.217,"cache_write_cost":0.39},"quantization":null,"context_size":1048576,"max_output_tokens":65535,"supported_features":["json_mode","reasoning","tools"],"metrics":{"throughput":127.19,"latency":24.51,"ttft":16.87,"cache_hit_rate":0.01,"uptime":1.0}}}},{"id":"gemma-3-27b-it","object":"model","created":1741779482,"owned_by":"Google","description":"Gemma 3 is a family of lightweight, multimodal models from Google, supporting text and image inputs, multilingual capabilities, and a 131K context window.","pricing":{"currency":"EUR","input_token":0.089,"output_token":0.268},"providers":["nebius"],"context_size":131000,"max_output_tokens":110000,"tags":["Instruct","Image","Tools","Reasoning"],"input_modalities":["text","image"],"output_modalities":["text"],"supported_features":["json_mode","reasoning"],"model_series":["gemma-3-27b"],"providers_details":{"nebius":{"pricing":{"currency":"EUR","input_token":0.089,"output_token":0.268},"quantization":"fp8","context_size":131000,"max_output_tokens":110000,"supported_features":["json_mode","reasoning"],"metrics":{"throughput":33.36,"latency":1.06,"ttft":0.25,"cache_hit_rate":0.57,"uptime":1.0}}}},{"id":"qwen2.5-vl-72b-instruct","object":"model","created":1737977882,"owned_by":"Alibaba Cloud","description":"Qwen2.5-VL is a powerful vision-language model with advanced capabilities in visual understanding, long video reasoning, and structured output generation.","pricing":{"currency":"EUR","input_token":0.91,"output_token":0.91},"providers":["ovh"],"context_size":32000,"max_output_tokens":32000,"tags":["Instruct","Image","Reasoning"],"input_modalities":["text","image"],"output_modalities":["text"],"supported_features":["json_mode","reasoning"],"model_series":["qwen-2-vl-72b"],"providers_details":{"ovh":{"pricing":{"currency":"EUR","input_token":0.91,"output_token":0.91},"quantization":"fp8","context_size":32000,"max_output_tokens":32000,"supported_features":["json_mode","reasoning"],"metrics":{"throughput":31.39,"latency":0.91,"ttft":0.2,"cache_hit_rate":null,"uptime":1.0}}}},{"id":"codestral-2508","object":"model","created":1736768282,"owned_by":"Mistral AI","description":"Codestral-2508 is a cutting-edge coding model optimized for low-latency, high-frequency tasks like FIM, code correction, and test generation.","pricing":{"currency":"EUR","input_token":0.33,"output_token":0.99,"cache_read_cost":0.033},"providers":["mistral"],"context_size":256000,"max_output_tokens":256000,"tags":["Instruct","Tools"],"input_modalities":["text"],"output_modalities":["text"],"supported_features":["json_mode","tools"],"model_series":["codestral"],"providers_details":{"mistral":{"pricing":{"currency":"EUR","input_token":0.33,"output_token":0.99,"cache_read_cost":0.033},"quantization":null,"context_size":256000,"max_output_tokens":256000,"supported_features":["json_mode","tools"],"metrics":{"throughput":144.0,"latency":1.06,"ttft":0.32,"cache_hit_rate":0.2,"uptime":1.0}}}},{"id":"llama-3.3-70b-instruct","object":"model","created":1733830682,"owned_by":"Meta","description":"With 70 billion parameters, this cutting-edge multilingual model excels in generating high-quality, safe, and helpful text across eight languages, making it stand out from other chat models.","pricing":{"currency":"EUR","input_token":0.65,"output_token":0.65},"providers":["scaleway","ovh","ionos","aki"],"context_size":131000,"max_output_tokens":131000,"tags":["Instruct","Tools","Reasoning"],"input_modalities":["text"],"output_modalities":["text"],"supported_features":["json_mode","reasoning","tools"],"model_series":["llama-3-70b"],"providers_details":{"scaleway":{"pricing":{"currency":"EUR","input_token":0.9,"output_token":0.9},"quantization":null,"context_size":131000,"max_output_tokens":16384,"supported_features":["json_mode","reasoning","tools"],"metrics":{"throughput":79.47,"latency":4.41,"ttft":0.5,"cache_hit_rate":null,"uptime":1.0}},"ovh":{"pricing":{"currency":"EUR","input_token":0.67,"output_token":0.67},"quantization":"fp8","context_size":131000,"max_output_tokens":131000,"supported_features":["json_mode","reasoning","tools"],"metrics":{"throughput":26.91,"latency":12.61,"ttft":3.95,"cache_hit_rate":null,"uptime":1.0}},"ionos":{"pricing":{"currency":"EUR","input_token":0.65,"output_token":0.65},"quantization":"int4","context_size":128000,"max_output_tokens":128000,"supported_features":["json_mode","reasoning","tools"],"metrics":{"throughput":47.1,"latency":7.81,"ttft":0.77,"cache_hit_rate":null,"uptime":1.0}},"aki":{"pricing":{"currency":"EUR","input_token":0.65,"output_token":0.65},"quantization":"fp8","context_size":128000,"max_output_tokens":32000,"supported_features":["json_mode","reasoning","tools"],"metrics":{"throughput":53.17,"latency":5.61,"ttft":0.42,"cache_hit_rate":0.9,"uptime":1.0}}}},{"id":"gpt-4o","object":"model","created":1732102682,"owned_by":"OpenAI","description":"GPT-4o is a transformative multimodal AI model that seamlessly fuses text, images, and audio to deliver exceptional speed, efficiency, and performance—at twice the speed and half the cost of GPT-4 Turbo.","pricing":{"currency":"EUR","input_token":2.387,"output_token":9.547,"cache_read_cost":1.194},"providers":["azure_sc","azure_spc"],"context_size":128000,"max_output_tokens":16000,"tags":["Instruct","Tools","Image","Reasoning","Document"],"input_modalities":["text","image","file"],"output_modalities":["text"],"supported_features":["reasoning","tools"],"model_series":["gpt-4o"],"providers_details":{"azure_sc":{"pricing":{"currency":"EUR","input_token":2.387,"output_token":9.547,"cache_read_cost":1.194},"quantization":null,"context_size":128000,"max_output_tokens":16000,"supported_features":["reasoning","tools"],"metrics":{"throughput":109.85,"latency":3.66,"ttft":2.96,"cache_hit_rate":0.46,"uptime":1.0}},"azure_spc":{"pricing":{"currency":"EUR","input_token":2.387,"output_token":9.547,"cache_read_cost":1.194},"quantization":null,"context_size":128000,"max_output_tokens":16000,"supported_features":["reasoning","tools"],"metrics":{"throughput":138.85,"latency":3.12,"ttft":1.36,"cache_hit_rate":0.92,"uptime":1.0}}}},{"id":"gpt-5-mini","object":"model","created":1732102682,"owned_by":"OpenAI","description":"GPT 5 mini is a compact, low-cost, and fast version of GPT-5, ideal for real-time agents and lightweight reasoning tasks.","pricing":{"currency":"EUR","input_token":0.25,"output_token":1.968,"cache_read_cost":0.05},"providers":["azure_sc"],"context_size":400000,"max_output_tokens":128000,"tags":["Instruct","Tools","Image","Reasoning"],"input_modalities":["text","image"],"output_modalities":["text"],"supported_features":["reasoning","tools"],"model_series":["gpt-5-mini"],"providers_details":{"azure_sc":{"pricing":{"currency":"EUR","input_token":0.25,"output_token":1.968,"cache_read_cost":0.05},"quantization":null,"context_size":400000,"max_output_tokens":128000,"supported_features":["reasoning","tools"],"metrics":{"throughput":184.15,"latency":3.86,"ttft":4.77,"cache_hit_rate":0.54,"uptime":1.0}}}},{"id":"gpt-5-nano","object":"model","created":1732102682,"owned_by":"OpenAI","description":"GPT 5 nano is the fastest, cheapest GPT-5 model, optimized for summarization, classification, and ultra-low latency tasks.","pricing":{"currency":"EUR","input_token":0.054,"output_token":0.394,"cache_read_cost":0.017},"providers":["azure_sc"],"context_size":400000,"max_output_tokens":128000,"tags":["Instruct","Tools","Image","Reasoning"],"input_modalities":["text","image"],"output_modalities":["text"],"supported_features":["reasoning","tools"],"model_series":["gpt-5-nano"],"providers_details":{"azure_sc":{"pricing":{"currency":"EUR","input_token":0.054,"output_token":0.394,"cache_read_cost":0.017},"quantization":null,"context_size":400000,"max_output_tokens":128000,"supported_features":["reasoning","tools"],"metrics":{"throughput":1472.33,"latency":15.57,"ttft":8.51,"cache_hit_rate":0.32,"uptime":1.0}}}},{"id":"pixtral-12b-2409","object":"model","created":1731176378,"owned_by":"Mistral AI","description":"Pixtral 2409 12B is a state-of-the-art multimodal model with 12B parameters and a 400M vision encoder, natively trained on interleaved text and image data. It excels in tasks spanning vision-language reasoning, instruction following, and pure text understanding, making it highly effective for real-world multimodal applications.","pricing":{"currency":"EUR","input_token":0.2,"output_token":0.2},"providers":["scaleway"],"context_size":128000,"max_output_tokens":4096,"tags":["Instruct","Image","Reasoning"],"input_modalities":["text","image"],"output_modalities":["text"],"supported_features":["json_mode","reasoning","tools"],"model_series":["pixtral-12b"],"providers_details":{"scaleway":{"pricing":{"currency":"EUR","input_token":0.2,"output_token":0.2},"quantization":null,"context_size":128000,"max_output_tokens":4096,"supported_features":["json_mode","reasoning","tools"],"metrics":{"throughput":94.61,"latency":0.84,"ttft":0.31,"cache_hit_rate":null,"uptime":1.0}}}},{"id":"hermes-4-405b","object":"model","created":1723549082,"owned_by":"NousResearch","description":"Hermes 4 405B is a frontier hybrid-mode reasoning model built on Llama 3.1, optimized for advanced logic, math, coding, and structured output generation.","pricing":{"currency":"EUR","input_token":0.894,"output_token":2.683},"providers":["nebius"],"context_size":128000,"max_output_tokens":128000,"tags":["Instruct","Tools"],"input_modalities":["text"],"output_modalities":["text"],"supported_features":["json_mode","tools"],"model_series":["hermes-4-405b"],"providers_details":{"nebius":{"pricing":{"currency":"EUR","input_token":0.894,"output_token":2.683},"quantization":"fp8","context_size":128000,"max_output_tokens":128000,"supported_features":["json_mode","tools"],"metrics":{"throughput":40.19,"latency":7.39,"ttft":0.44,"cache_hit_rate":0.02,"uptime":1.0}}}},{"id":"mistral-nemo-instruct-2407","object":"model","created":1723020793,"owned_by":"Mistral AI","description":"A 12B parameter, instruct-tuned language model by Mistral AI and NVIDIA, designed for advanced instruction following, multi-turn conversations, and generating text and code across multiple languages.","pricing":{"currency":"EUR","input_token":0.13,"output_token":0.13},"providers":["ovh","ionos"],"context_size":128000,"max_output_tokens":128000,"tags":["Instruct","Tools"],"input_modalities":["text"],"output_modalities":["text"],"supported_features":["json_mode","tools"],"model_series":["mistral-nemo"],"providers_details":{"ovh":{"pricing":{"currency":"EUR","input_token":0.13,"output_token":0.13},"quantization":"fp16","context_size":118000,"max_output_tokens":118000,"supported_features":["json_mode","tools"],"metrics":{"throughput":46.0,"latency":4.88,"ttft":0.39,"cache_hit_rate":null,"uptime":1.0}},"ionos":{"pricing":{"currency":"EUR","input_token":0.15,"output_token":0.15},"quantization":"fp8","context_size":128000,"max_output_tokens":128000,"supported_features":["json_mode","tools"],"metrics":{"throughput":48.41,"latency":7.75,"ttft":0.4,"cache_hit_rate":null,"uptime":1.0}}}},{"id":"gpt-4o-mini","object":"model","created":1721302682,"owned_by":"OpenAI","description":"GPT-4o mini is a lightweight, cost-efficient multimodal model optimized for real-time tasks and large-context applications, with support for text and vision.","pricing":{"currency":"EUR","input_token":0.143,"output_token":0.573,"cache_read_cost":0.073},"providers":["azure_sc","azure_spc"],"context_size":128000,"max_output_tokens":16000,"tags":["Instruct","Tools","Image","Reasoning","Document"],"input_modalities":["text","image","file"],"output_modalities":["text"],"supported_features":["reasoning","tools"],"model_series":["gpt-4o-mini"],"providers_details":{"azure_sc":{"pricing":{"currency":"EUR","input_token":0.143,"output_token":0.573,"cache_read_cost":0.073},"quantization":null,"context_size":128000,"max_output_tokens":16000,"supported_features":["reasoning","tools"],"metrics":{"throughput":73.83,"latency":3.4,"ttft":0.47,"cache_hit_rate":0.07,"uptime":1.0}},"azure_spc":{"pricing":{"currency":"EUR","input_token":0.143,"output_token":0.573,"cache_read_cost":0.073},"quantization":null,"context_size":128000,"max_output_tokens":16000,"supported_features":["reasoning","tools"],"metrics":{"throughput":72.53,"latency":3.3,"ttft":0.66,"cache_hit_rate":0.09,"uptime":1.0}}}},{"id":"llama-3.1-8b-instruct","object":"model","created":1712652299,"owned_by":"Meta","description":"Optimized for dialogue, this LLM by Meta outperforms other open-source chat models in benchmarks while prioritizing helpfulness and safety.","pricing":{"currency":"EUR","input_token":0.15,"output_token":0.15},"providers":["ionos","aki"],"context_size":128000,"max_output_tokens":128000,"tags":["Instruct","Tools","Reasoning"],"input_modalities":["text"],"output_modalities":["text"],"supported_features":["json_mode","reasoning","tools"],"model_series":["llama-3-8b"],"providers_details":{"ionos":{"pricing":{"currency":"EUR","input_token":0.15,"output_token":0.15},"quantization":"fp8","context_size":128000,"max_output_tokens":128000,"supported_features":["json_mode","reasoning","tools"],"metrics":{"throughput":63.95,"latency":4.93,"ttft":0.2,"cache_hit_rate":null,"uptime":1.0}},"aki":{"pricing":{"currency":"EUR","input_token":0.15,"output_token":0.15},"quantization":"fp8","context_size":65536,"max_output_tokens":16384,"supported_features":["json_mode","reasoning","tools"],"metrics":{"throughput":93.46,"latency":3.35,"ttft":0.42,"cache_hit_rate":0.54,"uptime":1.0}}}},{"id":"mixtral-8x7B-instruct-v0.1","object":"model","created":1702294682,"owned_by":"Mistral AI","description":"Mixtral-8x7B is a sparse mixture-of-experts model with open weights, delivering fast, high-performance inference that outperforms Llama 2 70B and rivals GPT-3.5.","pricing":{"currency":"EUR","input_token":0.438,"output_token":0.68},"providers":["amazon_ireland","amazon_paris"],"context_size":32000,"max_output_tokens":4096,"tags":["Instruct","Reasoning","Document"],"input_modalities":["text","file"],"output_modalities":["text"],"supported_features":["json_mode","reasoning"],"model_series":["mixtral-8x7b"],"providers_details":{"amazon_ireland":{"pricing":{"currency":"EUR","input_token":0.438,"output_token":0.68},"quantization":null,"context_size":32000,"max_output_tokens":4096,"supported_features":["json_mode","reasoning"],"metrics":{"throughput":112.79,"latency":3.61,"ttft":0.24,"cache_hit_rate":null,"uptime":1.0}},"amazon_paris":{"pricing":{"currency":"EUR","input_token":0.528,"output_token":0.814},"quantization":null,"context_size":32000,"max_output_tokens":4096,"supported_features":["json_mode","reasoning"],"metrics":{"throughput":112.35,"latency":3.36,"ttft":0.21,"cache_hit_rate":null,"uptime":1.0}}}}]}