Integrations / Provider Guides
Z.ai
Connect Aurora to Z.ai's GLM models for cheap, capable Chinese-language chat.
Browse docs
Use this provider to route to Z.ai's hosted GLM models. Z.ai is OpenAI-compatible; Aurora translates between formats with no special handling.
Activate
text
ZAI_API_KEY=...Restart Aurora. The zai provider appears at /v1/models with the current GLM catalog.
Default models
Override with ZAI_MODELS=glm-4.6,glm-4.5.
Multiple instances
text
ZAI_PROD_API_KEY=...
ZAI_STAGING_API_KEY=...Aurora exposes them as zai-prod and zai-staging.
Custom base URL
text
ZAI_BASE_URL=https://api.z.ai/api/paas/v4Use this for the Z.ai general-API endpoint. Z.ai also offers a coding-specific endpoint — set ZAI_BASE_URL to its URL to use it.
Gotchas
- Reasoning models — Z.ai returns reasoning content as
reasoningblocks. Aurora preserves them in the OpenAI-shaped response. - Tool use — full support, no quirks.
- Prompt caching — not supported by Z.ai. Use Aurora's semantic cache if you need replay.
- Function calling — full support.
Sample request
text
curl http://localhost:8080/v1/chat/completions \
-H "Content-Type: application/json" \
-H "Authorization: Bearer $AURORA_MASTER_KEY" \
-d '{
"model": "zai/glm-4.6",
"messages": [{"role": "user", "content": "你好"}]
}'