Z.AI Model Config

created: 2024-06-24
status: active

Z.AI Model Config

Hermes model hierarchy and fallback chain for Z.AI provider.

Current Hierarchy

  1. glm-5.2 (primary)
  2. glm-5.1 (fallback 1)
  3. glm-5-turbo (fallback 2)
  4. glm-4.7 (fallback 3)
  5. glm-4.5-air (fallback 4)
  6. gemini (fallback 5)
  7. cerebras (fallback 6)
  8. groq (fallback 7)

Known Quirks

  • GLM 5.2 reasoning overhead: Uses 1000-2000 reasoning_tokens internally before output. max_tokens must be 8000+ for essay-length content or output comes back empty.
  • Vision fallback: glm-4.5v
  • Bare model names only: No [1m] suffix

API Endpoint

https://api.z.ai/api/coding/paas/v4

↩ Linked from