inference-providers

Z.ai

first_party

Docs Console Pricing

Getting credentials

api-key

api_key via header (header Authorization: Bearer)

Environment: ZAI_API_KEY

z.ai console → API Keys. Key format id.secret; JWT HS256 derivation also supported.

Credentials docs

Endpoints

ID	URL	Protocol	Auth
chat-completions	`https://api.z.ai/api/paas/v4/chat/completions`	openai-chat	any

API surfaces

textstreaming

Quirks

clear_thinking: false preserves reasoning_content across turns (Preserved Thinking); disabled by default on this endpoint. (docs)
GLM-5.3-Flash is the first multimodal GLM-5-series model: chat-completions accepts image/video/file inputs alongside text. (docs)

Models offered

Model	Wire ID	Endpoint	Status	Reasoning
zai/glm-4-6	`glm-4.6`	chat-completions	ga	toggle · default on · returns reasoning_content · round-trips reasoning_content
zai/glm-5-1	`glm-5.1`	chat-completions	ga	toggle · default on · returns reasoning_content · round-trips reasoning_content
zai/glm-5-3	`glm-5.3`	chat-completions	ga	always_on · default on · mandatory · returns reasoning_content · round-trips reasoning_content
zai/glm-5-3-flash	`glm-5.3-flash`	chat-completions	ga	toggle · default on · mandatory · returns reasoning_content