OpenAI-compatible chat/completion proxy backed by https://chat.z.ai/.

• Supports POST /v1/chat/completions
• Supports POST /v1/responses
• Creates a fresh upstream chat for every request
• Preserves reasoning output separately from final answer text
• Reuses ZAI_SESSION_TOKEN directly or refreshes it from ZAI_JWT

• Python 3.12+
• uv
• One of:
  • ZAI_JWT
  • ZAI_SESSION_TOKEN

export ZAI_JWT='your-jwt'
uv run python -m zai2api

Or with the installed script:

export ZAI_JWT='your-jwt'
uv run zai2api

Default bind address is 0.0.0.0:8000.

• ZAI_JWT: preferred auth source; used to fetch a fresh session token
• ZAI_SESSION_TOKEN: optional direct session token reuse
• DEFAULT_MODEL: defaults to glm-5
• Available public model ids: glm-5, glm-5.1, glm-5-turbo and their -nothinking variants
• HOST: defaults to 0.0.0.0
• PORT: defaults to 8000
• LOG_LEVEL: defaults to info
• REQUEST_TIMEOUT: defaults to 120

curl http://127.0.0.1:8000/v1/chat/completions \
  -H 'content-type: application/json' \
  -d '{
    "model": "glm-5",
    "messages": [
      {"role": "system", "content": "Be concise."},
      {"role": "user", "content": "Say hello."}
    ]
  }'

curl http://127.0.0.1:8000/v1/responses \
  -H 'content-type: application/json' \
  -d '{
    "model": "glm-5",
    "input": "Say hello."
  }'