[DOCS]

Models

GET /v1/models returns the live models in the stock OpenAI list format. Model ids are used verbatim in chat/completions.

GET /v1/models
curl https://blazeinference.com/v1/models

{
  "object": "list",
  "data": [
    {
      "id": "deepseek-ai/deepseek-v4-pro",
      "object": "model",
      "created": 1745971200,
      "owned_by": "deepseek-ai"
    },
    {
      "id": "deepseek-ai/deepseek-v4-flash",
      "object": "model",
      "created": 1747699200,
      "owned_by": "deepseek-ai"
    },
    {
      "id": "z-ai/glm-5.2",
      "object": "model",
      "created": 1781827200,
      "owned_by": "z-ai"
    },
    {
      "id": "moonshotai/kimi-k2.6",
      "object": "model",
      "created": 1768435200,
      "owned_by": "moonshotai"
    }
  ]
}

Available in the demo

MODEL IDNOTES
deepseek-ai/deepseek-v4-pro1.6T MoE. Reasoning off by default; opt in via chat_template_kwargs.
deepseek-ai/deepseek-v4-flashV4 speed tier. Same optional reasoning, plus reasoning_effort control.
z-ai/glm-5.2Fast, strong coding/agentic model. MIT-licensed weights.
moonshotai/kimi-k2.6Moonshot's K-series workhorse. 256K context.

Coming soon

Upcoming models (Kimi K3 next) appear on the models page until they are available; the API lists only models that can be called right now. Requesting a not-yet-available model returns model_not_available.