{"data":[{"id":"kimi-k3","type":"llm","section":"models","name":"Kimi K3","maker":"Moonshot AI","category":"reasoning","provider":"GPU enclave","summary":"The largest model listed. Long multi-step reasoning, agents and code.","modality":"text+vision","context_length":200000,"min_output_tokens":512,"input_usd_per_mtok":3,"output_usd_per_mtok":15,"usd_per_image":null,"unit":null,"usd_per_unit":null,"endpoint":"/v1/chat/completions","confidential":true},{"id":"glm-5.3","type":"llm","section":"models","name":"GLM 5.3","maker":"Z.ai","category":"reasoning","provider":"GPU enclave","summary":"Software engineering and long-horizon agent work.","modality":"text","context_length":200000,"min_output_tokens":512,"input_usd_per_mtok":1.4,"output_usd_per_mtok":4.4,"usd_per_image":null,"unit":null,"usd_per_unit":null,"endpoint":"/v1/chat/completions","confidential":true},{"id":"glm-5.2","type":"llm","section":"models","name":"GLM 5.2","maker":"Z.ai","category":"reasoning","provider":"GPU enclave","summary":"Long-horizon tasks over very large inputs.","modality":"text","context_length":200000,"min_output_tokens":512,"input_usd_per_mtok":1.26,"output_usd_per_mtok":3,"usd_per_image":null,"unit":null,"usd_per_unit":null,"endpoint":"/v1/chat/completions","confidential":true},{"id":"glm-5.1","type":"llm","section":"models","name":"GLM 5.1","maker":"Z.ai","category":"reasoning","provider":"GPU enclave","summary":"A coding-focused model for multi-step engineering work.","modality":"text","context_length":200000,"min_output_tokens":512,"input_usd_per_mtok":1.21,"output_usd_per_mtok":4.2,"usd_per_image":null,"unit":null,"usd_per_unit":null,"endpoint":"/v1/chat/completions","confidential":true},{"id":"kimi-k2.6","type":"llm","section":"models","name":"Kimi K2.6","maker":"Moonshot AI","category":"reasoning","provider":"GPU enclave","summary":"Multimodal reasoning for coding and interface work.","modality":"text+vision","context_length":200000,"min_output_tokens":512,"input_usd_per_mtok":1.09,"output_usd_per_mtok":4.6,"usd_per_image":null,"unit":null,"usd_per_unit":null,"endpoint":"/v1/chat/completions","confidential":true},{"id":"qwen3.5-397b","type":"llm","section":"models","name":"Qwen3.5 397B","maker":"Qwen","category":"reasoning","provider":"GPU enclave","summary":"Vision and language in one model, at a lower price per token.","modality":"text+vision","context_length":200000,"min_output_tokens":512,"input_usd_per_mtok":0.55,"output_usd_per_mtok":3.5,"usd_per_image":null,"unit":null,"usd_per_unit":null,"endpoint":"/v1/chat/completions","confidential":true},{"id":"deepseek-v4-flash","type":"llm","section":"models","name":"DeepSeek V4 Flash","maker":"DeepSeek","category":"general","provider":"GPU enclave","summary":"Everyday coding, reasoning and agent tasks.","modality":"text","context_length":200000,"min_output_tokens":512,"input_usd_per_mtok":0.44,"output_usd_per_mtok":1.32,"usd_per_image":null,"unit":null,"usd_per_unit":null,"endpoint":"/v1/chat/completions","confidential":true},{"id":"deepseek-v3.2","type":"llm","section":"models","name":"DeepSeek V3.2","maker":"DeepSeek","category":"general","provider":"GPU enclave","summary":"One flat price for input and output.","modality":"text","context_length":163840,"min_output_tokens":512,"input_usd_per_mtok":1,"output_usd_per_mtok":1,"usd_per_image":null,"unit":null,"usd_per_unit":null,"endpoint":"/v1/chat/completions","confidential":true},{"id":"qwen3.8-27b","type":"llm","section":"models","name":"Qwen3.8 27B","maker":"Qwen","category":"general","provider":"GPU enclave","summary":"A compact vision-language model for coding and documents.","modality":"text+vision","context_length":200000,"min_output_tokens":512,"input_usd_per_mtok":0.2,"output_usd_per_mtok":2.5,"usd_per_image":null,"unit":null,"usd_per_unit":null,"endpoint":"/v1/chat/completions","confidential":true},{"id":"qwen3.6-27b","type":"llm","section":"models","name":"Qwen3.6 27B","maker":"Qwen","category":"general","provider":"GPU enclave","summary":"A dense 27B model that reads text, images and video.","modality":"text+vision","context_length":200000,"min_output_tokens":512,"input_usd_per_mtok":0.32,"output_usd_per_mtok":3.25,"usd_per_image":null,"unit":null,"usd_per_unit":null,"endpoint":"/v1/chat/completions","confidential":true},{"id":"qwen3.5-27b","type":"llm","section":"models","name":"Qwen3.5 27B","maker":"Qwen","category":"general","provider":"GPU enclave","summary":"Vision and language with quick responses.","modality":"text+vision","context_length":200000,"min_output_tokens":512,"input_usd_per_mtok":0.3,"output_usd_per_mtok":2.4,"usd_per_image":null,"unit":null,"usd_per_unit":null,"endpoint":"/v1/chat/completions","confidential":true},{"id":"llama-3.3-70b","type":"llm","section":"models","name":"Llama 3.3 70B","maker":"Meta","category":"general","provider":"GPU enclave","summary":"A widely used instruction model with tool calling and JSON output.","modality":"text","context_length":131072,"min_output_tokens":1,"input_usd_per_mtok":2,"output_usd_per_mtok":2,"usd_per_image":null,"unit":null,"usd_per_unit":null,"endpoint":"/v1/chat/completions","confidential":true},{"id":"gpt-oss-120b","type":"llm","section":"models","name":"GPT OSS 120B","maker":"OpenAI","category":"general","provider":"GPU enclave","summary":"OpenAI's open-weight model. Low cost for general agent reasoning.","modality":"text","context_length":131072,"min_output_tokens":512,"input_usd_per_mtok":0.15,"output_usd_per_mtok":0.6,"usd_per_image":null,"unit":null,"usd_per_unit":null,"endpoint":"/v1/chat/completions","confidential":true},{"id":"glm-5.3-flash","type":"llm","section":"models","name":"GLM 5.3 Flash","maker":"Z.ai","category":"fast","provider":"GPU enclave","summary":"Built for high-volume agent tasks.","modality":"text","context_length":200000,"min_output_tokens":512,"input_usd_per_mtok":0.15,"output_usd_per_mtok":0.5,"usd_per_image":null,"unit":null,"usd_per_unit":null,"endpoint":"/v1/chat/completions","confidential":true},{"id":"qwen3.6-35b","type":"llm","section":"models","name":"Qwen3.6 35B","maker":"Qwen","category":"fast","provider":"GPU enclave","summary":"Only 3B of its 35B parameters run per token, so it stays quick and cheap.","modality":"text+vision","context_length":200000,"min_output_tokens":512,"input_usd_per_mtok":0.2,"output_usd_per_mtok":1.27,"usd_per_image":null,"unit":null,"usd_per_unit":null,"endpoint":"/v1/chat/completions","confidential":true},{"id":"qwen3-vl-30b","type":"llm","section":"models","name":"Qwen3 VL 30B","maker":"Qwen","category":"fast","provider":"GPU enclave","summary":"Reads images and documents at a low price per token.","modality":"text+vision","context_length":128000,"min_output_tokens":1,"input_usd_per_mtok":0.2,"output_usd_per_mtok":0.7,"usd_per_image":null,"unit":null,"usd_per_unit":null,"endpoint":"/v1/chat/completions","confidential":true},{"id":"qwen3-32b","type":"llm","section":"models","name":"Qwen3 32B","maker":"Qwen","category":"fast","provider":"GPU enclave","summary":"A small reasoning model for short tasks.","modality":"text","context_length":40960,"min_output_tokens":512,"input_usd_per_mtok":0.12,"output_usd_per_mtok":0.5,"usd_per_image":null,"unit":null,"usd_per_unit":null,"endpoint":"/v1/chat/completions","confidential":true},{"id":"qwen2.5-7b","type":"llm","section":"models","name":"Qwen2.5 7B","maker":"Qwen","category":"fast","provider":"GPU enclave","summary":"The cheapest item listed. Simple classification, extraction and rewriting.","modality":"text","context_length":32768,"min_output_tokens":1,"input_usd_per_mtok":0.1,"output_usd_per_mtok":0.2,"usd_per_image":null,"unit":null,"usd_per_unit":null,"endpoint":"/v1/chat/completions","confidential":true},{"id":"gemma-4-31b","type":"llm","section":"models","name":"Gemma 4 31B","maker":"Google","category":"fast","provider":"GPU enclave","summary":"Answers without a thinking phase, so short replies stay short and cheap.","modality":"text+vision","context_length":200000,"min_output_tokens":1,"input_usd_per_mtok":0.15,"output_usd_per_mtok":0.46,"usd_per_image":null,"unit":null,"usd_per_unit":null,"endpoint":"/v1/chat/completions","confidential":true},{"id":"qwen3.8-27b-uncensored","type":"llm","section":"models","name":"Qwen3.8 27B Uncensored","maker":"Qwen","category":"uncensored","provider":"GPU enclave","summary":"A community fine-tune of Qwen3.8 27B that does not refuse.","modality":"text+vision","context_length":200000,"min_output_tokens":512,"input_usd_per_mtok":0.3,"output_usd_per_mtok":1.5,"usd_per_image":null,"unit":null,"usd_per_unit":null,"endpoint":"/v1/chat/completions","confidential":true},{"id":"gemma-4-26b-uncensored","type":"llm","section":"models","name":"Gemma 4 26B Uncensored","maker":"Google","category":"uncensored","provider":"GPU enclave","summary":"A community fine-tune of Gemma 4 26B that does not refuse.","modality":"text+vision","context_length":200000,"min_output_tokens":1,"input_usd_per_mtok":0.15,"output_usd_per_mtok":0.7,"usd_per_image":null,"unit":null,"usd_per_unit":null,"endpoint":"/v1/chat/completions","confidential":true},{"id":"flux-schnell","type":"image","section":"media","name":"FLUX Schnell","maker":"Black Forest Labs","category":"image","provider":"Cloud GPU","summary":"The cheapest image item. A picture in about a second.","modality":"image","context_length":0,"min_output_tokens":1,"input_usd_per_mtok":0,"output_usd_per_mtok":0,"usd_per_image":0.003,"unit":"image","usd_per_unit":0.003,"endpoint":"/v1/images/generations","confidential":false},{"id":"flux-dev","type":"image","section":"media","name":"FLUX Dev","maker":"Black Forest Labs","category":"image","provider":"Cloud GPU","summary":"More detail and closer prompt following than Schnell.","modality":"image","context_length":0,"min_output_tokens":1,"input_usd_per_mtok":0,"output_usd_per_mtok":0,"usd_per_image":0.025,"unit":"image","usd_per_unit":0.025,"endpoint":"/v1/images/generations","confidential":false},{"id":"flux-1.1-pro","type":"image","section":"media","name":"FLUX 1.1 Pro","maker":"Black Forest Labs","category":"image","provider":"Cloud GPU","summary":"High image quality and varied results from one prompt.","modality":"image","context_length":0,"min_output_tokens":1,"input_usd_per_mtok":0,"output_usd_per_mtok":0,"usd_per_image":0.04,"unit":"image","usd_per_unit":0.04,"endpoint":"/v1/images/generations","confidential":false},{"id":"imagen-4-fast","type":"image","section":"media","name":"Imagen 4 Fast","maker":"Google","category":"image","provider":"Cloud GPU","summary":"Google's image model, in its quicker and cheaper form.","modality":"image","context_length":0,"min_output_tokens":1,"input_usd_per_mtok":0,"output_usd_per_mtok":0,"usd_per_image":0.02,"unit":"image","usd_per_unit":0.02,"endpoint":"/v1/images/generations","confidential":false},{"id":"imagen-4","type":"image","section":"media","name":"Imagen 4","maker":"Google","category":"image","provider":"Cloud GPU","summary":"Photorealism and fine detail.","modality":"image","context_length":0,"min_output_tokens":1,"input_usd_per_mtok":0,"output_usd_per_mtok":0,"usd_per_image":0.04,"unit":"image","usd_per_unit":0.04,"endpoint":"/v1/images/generations","confidential":false},{"id":"nano-banana","type":"image","section":"media","name":"Nano Banana","maker":"Google","category":"image","provider":"Cloud GPU","summary":"Google's Gemini image model. Strong at following long instructions.","modality":"image","context_length":0,"min_output_tokens":1,"input_usd_per_mtok":0,"output_usd_per_mtok":0,"usd_per_image":0.039,"unit":"image","usd_per_unit":0.039,"endpoint":"/v1/images/generations","confidential":false},{"id":"ideogram-v3-turbo","type":"image","section":"media","name":"Ideogram V3 Turbo","maker":"Ideogram","category":"image","provider":"Cloud GPU","summary":"Known for getting text inside an image right.","modality":"image","context_length":0,"min_output_tokens":1,"input_usd_per_mtok":0,"output_usd_per_mtok":0,"usd_per_image":0.03,"unit":"image","usd_per_unit":0.03,"endpoint":"/v1/images/generations","confidential":false},{"id":"seedream-4.5","type":"image","section":"media","name":"Seedream 4.5","maker":"ByteDance","category":"image","provider":"Cloud GPU","summary":"Cinematic, high-resolution pictures.","modality":"image","context_length":0,"min_output_tokens":1,"input_usd_per_mtok":0,"output_usd_per_mtok":0,"usd_per_image":0.04,"unit":"image","usd_per_unit":0.04,"endpoint":"/v1/images/generations","confidential":false},{"id":"qwen-image","type":"image","section":"media","name":"Qwen Image","maker":"Qwen","category":"image","provider":"Cloud GPU","summary":"Renders text well, in English and Chinese.","modality":"image","context_length":0,"min_output_tokens":1,"input_usd_per_mtok":0,"output_usd_per_mtok":0,"usd_per_image":0.025,"unit":"image","usd_per_unit":0.025,"endpoint":"/v1/images/generations","confidential":false},{"id":"recraft-v3","type":"image","section":"media","name":"Recraft V3","maker":"Recraft","category":"image","provider":"Cloud GPU","summary":"Design-oriented: illustrations, posters and long text in images.","modality":"image","context_length":0,"min_output_tokens":1,"input_usd_per_mtok":0,"output_usd_per_mtok":0,"usd_per_image":0.04,"unit":"image","usd_per_unit":0.04,"endpoint":"/v1/images/generations","confidential":false}]}