Models

Use GET /models to discover the exact model IDs accepted by the public API. Only models returned by this endpoint can be used in /chat/completions.

curl https://glm5.app/api/v1/models \
  -H "Authorization: Bearer $GLM5_API_KEY"

{
  "object": "list",
  "data": [
    {
      "id": "glm-5.3-flash",
      "object": "model",
      "created": 0,
      "owned_by": "glm5"
    }
  ]
}

Model List

Rates are shown in credits per one million input or output tokens. API calls spend the same GLM 5 credit balance used by Chat. Dollar equivalents are only a secondary reference. See Billing for the credit conversion, reservation, settlement, and refund behavior.

GET /models does not publish synthetic context_length or max_completion_tokens values. GLM 5 does not impose one global token window; context and output limits can vary by model.

Model IDs

Use the exact model IDs returned by GET /models. Send one of these values:

If a model ID is not returned by GET /models, the API returns 404 model_not_found.

GPT-6 Astra

gpt-6-astra is the public API ID for OpenAI GPT-6 Astra. Use it for frontier reasoning, coding, document work, and long-context tasks. It supports image input and tool calling through the same OpenAI-compatible Chat Completions endpoint.

The public API rate is $50 / 1M input tokens and $250 / 1M output tokens. Requests whose prompt reaches 272,000 input tokens are billed for the whole request at the long-context rate of $100 / 1M input and $375 / 1M output, mirroring the upstream tier. The threshold applies to the request's input tokens, and the higher rate then applies to input and output alike.

{
  "model": "gpt-6-astra",
  "messages": [
    {
      "role": "user",
      "content": "Read this incident timeline and derive the root cause with the evidence for each step."
    }
  ],
  "max_completion_tokens": 1600
}

Claude Fable 5.1

claude-fable-5.1 is the public API ID for Anthropic Claude Fable 5.1. It is optimized for long-horizon reasoning, coding, document work, and agent workflows. It supports image input and tool calling through the same OpenAI-compatible Chat Completions endpoint.

The public API rate is $50 / 1M input tokens and $250 / 1M output tokens. The model uses the shared credit balance and the standard API reservation and settlement flow.

{
  "model": "claude-fable-5.1",
  "messages": [
    {
      "role": "user",
      "content": "Review this repository migration plan and identify hidden long-horizon risks."
    }
  ],
  "max_completion_tokens": 1600
}

GLM 5.3 Flash (Ox Alpha)

GLM 5.3 Flash is the official model name; Ox Alpha was its preview codename. Use glm-5.3-flash as the default model ID for new API integrations. It accepts text, image, and video input, returns text, and supports function calling and structured outputs. Use it for responsive coding, agent workflows, and long-context tasks when you want the lowest-cost current GLM API entry point.

{
  "model": "glm-5.3-flash",
  "messages": [
    {
      "role": "user",
      "content": "Review this deployment plan and identify the highest-risk rollback gaps."
    }
  ],
  "max_completion_tokens": 1200
}

For migration only, the legacy input aliases ox-alpha and stealth/ox-alpha are accepted and resolve to glm-5.3-flash. They are not returned by GET /models and must not be used in new integrations.

GLM-5.3

glm-5.3 remains the flagship GLM option for complex coding, structured reasoning, agent workflows, function calling, and longer technical conversations.

{
  "model": "glm-5.3",
  "messages": [
    {
      "role": "user",
      "content": "Review this deployment plan for missing rollback steps."
    }
  ],
  "max_completion_tokens": 1200
}

Use it when you want the current GLM API entry point for new integrations.

GLM-5.2

glm-5.2 remains available as a stable GLM model ID for existing integrations, coding, reasoning, and function-calling workflows.

{
  "model": "glm-5.2",
  "messages": [
    {
      "role": "user",
      "content": "Summarize this incident report and list the next actions."
    }
  ],
  "max_completion_tokens": 1000
}

GLM-5

glm-5 is a lower-cost GLM option for general chat, summarization, structured answers, and lighter application workflows. It supports the same text endpoint, streaming, and function-calling shape as glm-5.3.

{
  "model": "glm-5",
  "messages": [
    {
      "role": "user",
      "content": "Turn these notes into a concise customer support reply."
    }
  ],
  "max_completion_tokens": 700
}

Kimi K3

kimi-k3 is a premium alternative model exposed through the same Chat Completions API. Use it when you want to compare behavior against GLM models for reasoning-heavy or high-value text workflows.

{
  "model": "kimi-k3",
  "messages": [
    {
      "role": "user",
      "content": "Compare these two architecture options and choose one."
    }
  ],
  "max_completion_tokens": 1500
}

Kimi K2

kimi-k2 is a cost-efficient alternative for general chat, summaries, extraction, and tool-capable workflows.

{
  "model": "kimi-k2",
  "messages": [
    {
      "role": "user",
      "content": "Summarize the trade-offs in this architecture."
    }
  ],
  "max_completion_tokens": 800
}

DeepSeek R1

deepseek-r1 is available for reasoning-oriented text generation. It does not support function calling in the GLM 5 public API. If you send tools with this model, the API returns 400 unsupported_parameter.

{
  "model": "deepseek-r1",
  "messages": [
    {
      "role": "user",
      "content": "Reason through this production incident timeline."
    }
  ],
  "max_completion_tokens": 1200
}

DeepSeek V4 Pro

deepseek-v4-pro exposes DeepSeek V4 Pro through the GLM 5 public API. Use it for higher-value reasoning, coding, extraction, and tool-capable text workflows.

{
  "model": "deepseek-v4-pro",
  "messages": [
    {
      "role": "user",
      "content": "Find the risky assumptions in this migration plan."
    }
  ],
  "max_completion_tokens": 1200
}

DeepSeek V4 Flash

deepseek-v4-flash is the cheaper DeepSeek V4 option for fast everyday chat, summaries, and tool-capable application workflows.

{
  "model": "deepseek-v4-flash",
  "messages": [
    {
      "role": "user",
      "content": "Summarize these customer notes into action items."
    }
  ],
  "max_completion_tokens": 800
}

Model Not Found

Unknown model IDs return:

{
  "error": {
    "message": "The requested model was not found.",
    "type": "invalid_request_error",
    "code": "model_not_found",
    "param": null
  }
}

The HTTP status is 404.