Z.AI (GLM)

[official website]

GLM models can be added to Tresor via API key authentication. The general API and GLM Coding Plan use different endpoints. Start with the general API configuration below, or use the separate Coding Plan profile later on this page.

Web UI

Web UI is more intuitive, but requires more manual operations.

In the "Downstreams" tab:

• Click + Add Provider if Z.AI is not already listed.
• Enter Z.AI or any name you like.
• Under "API Formats", select only OpenAI for the general API.
• Under "Authentication", select API key and enter a key from the API Keys management page linked in Z.AI's quick-start.
• Under "API Host", enter https://api.z.ai.
• Click Advanced and set the OpenAI per-format request path to /api/paas/v4/chat/completions.
• Under "Models", click Fetch Models. If discovery is unavailable, manually add the exact GLM model IDs shown in your console.

config.yaml

Editing YAML is less intuitive, but requires fewer manual operations.

Replace the API key and model ID placeholders with values from your Z.AI console.

~/.config/tresor/config.yaml

# Other settings omitted. Add this entry to your existing downstreams list.
downstreams:
  - id: zai
    name: Z.AI
    base_url: https://api.z.ai
    api_formats:
      - openai
    format_paths:
      openai: /api/paas/v4/chat/completions
    output_model_ids:
      - YOUR-GLM-MODEL-ID
    is_enabled: true
    auth:
      type: api_key
      api_key: YOUR-ZAI-API-KEY

Restart Tresor to apply the YAML changes.

GLM Coding Plan

Create a separate downstream for the Coding Plan. In the Web UI, select OpenAI, OpenAI Responses, and Anthropic, use https://api.z.ai as the API host, and set these request paths under Advanced:

Format	Per-format request path
OpenAI	/api/coding/paas/v4/chat/completions
OpenAI Responses	/api/v1/responses
Anthropic	/api/anthropic/v1/messages

Use the credentials issued for your Coding Plan, not an unrelated general API key. The equivalent YAML entry is:

downstreams:
  - id: zai-coding
    name: Z.AI Coding Plan
    base_url: https://api.z.ai
    api_formats:
      - openai
      - openai_responses
      - anthropic
    format_paths:
      openai: /api/coding/paas/v4/chat/completions
      openai_responses: /api/v1/responses
      anthropic: /api/anthropic/v1/messages
    output_model_ids:
      - YOUR-CODING-PLAN-MODEL-ID
    is_enabled: true
    auth:
      type: api_key
      api_key: YOUR-ZAI-CODING-PLAN-API-KEY

Merge this entry into your existing downstreams list and restart Tresor. Using the wrong endpoint can prevent use of Coding Plan quota; team credentials may also differ from ordinary Z.AI keys.

Done! Clients accessing Tresor can now use your configured GLM models.

Next step

• Configure your LLM Apps