China AI Hub China AI Hub

API / zai

zai API

Provider: zhipu-ai · Type: official

zai is an API by zhipu-ai, served at https://api.z.ai/api/paas/v4/chat/completions for official models, exposing GLM-5.2, GLM-5.3-Flash, GLM-5.3-FlashX, GLM-5.3.

zai API
Image: AI-generated illustration (Seedream)
Last verified: · Review cadence: 90 days · Next review:
Key facts for zai API
Key factValue
Endpointhttps://api.z.ai/api/paas/v4/chat/completions
AuthenticationBearer API key
StreamingYes
Function callingYes
Tool callingYes
Structured outputUnknown
VisionYes
AudioUnknown
Regionsinternational, china
Cloud providersZ.ai, BigModel
Documentationhttps://docs.z.ai/

Context limits

Context limits for zai API
ModelInput limitOutput limit
glm-5.3 1,048,576 131,072
glm-5.3-flash 1,048,576 131,072

What it is. The Z.ai API is Zhipu AI’s international platform serving the GLM family through an OpenAI-compatible Chat Completions endpoint, with Responses and Anthropic Messages protocols on GLM-5.3. Why it matters. It is the international distribution path for GLM-5.3 and GLM-5.3-Flash, the latter at a budget-tier $0.15/$0.50 per 1M tokens. Key characteristics. Three protocol surfaces (Chat Completions, Responses, Anthropic Messages) plus a dedicated coding-plan endpoint; tool_stream: true for tool streaming (recommended for Flash models); 1M-token context. What a professional should know. The China platform (BigModel, open.bigmodel.cn) is operated separately, and the rate-limit documentation page was not located (docs.z.ai paths 404) — a gap to verify before production use.

The Z.ai API (Zhipu’s international platform) serves the GLM family via an OpenAI-compatible Chat Completions endpoint at https://api.z.ai/api/paas/v4/chat/completions. GLM-5.3 additionally supports the OpenAI Responses protocol (base https://api.z.ai/api/v1) and the Anthropic Messages protocol (base https://api.z.ai/api/anthropic); the GLM Coding Plan uses https://api.z.ai/api/coding/paas/v4.

Authentication is a Bearer API key. Streaming is supported, with tool_stream: true for tool-streaming (recommended for Flash models). The China platform (BigModel, open.bigmodel.cn) is operated separately.

Why it matters

The Z.ai API matters as the multi-protocol surface for open-weight GLM models: it is one of the few tracked APIs exposing OpenAI, Anthropic and a coding-plan endpoint from a single provider, which lowers migration cost for teams that want both general and coding access to the same GLM-5.3 family. The cost trade-off is favorable at the flash tier ($0.15/$0.50) and moderate at the flagship ($1.40/$4.40), while the documented gap — no located rate-limit page — is an operational uncertainty. China AI Hub analysis indicates the API’s structural role is the international, protocol-flexible entry to Zhipu’s open model family, with the BigModel split meaning China and international are managed as separate surfaces.

Labels used above: Official fact (from Z.ai Quick Start and GLM-5.3 docs), Vendor-reported claim (pricing and capability statements by Zhipu AI), and China AI Hub analysis (our synthesis, always introduced as such).

Sources

Data interpretation

Evidence types for zai API
FieldValueEvidence type
Endpointhttps://api.z.ai/api/paas/v4/chat/completionsOfficial
AuthenticationBearer API keyOfficial
CapabilitiesStreaming, Function calling, Tool calling, VisionOfficial
Models served4 models in the databaseChina AI Hub analysis

Evidence types: Official = vendor documentation or official pages. China AI Hub analysis = derived from the database itself. See the sourcing policy.

What is the zai API endpoint?

The zai API endpoint is https://api.z.ai/api/paas/v4/chat/completions.

How do you authenticate with the zai API?

Bearer API key.

What are the context limits of the zai API?

glm-5.3: 1,048,576 in / 131,072 out; glm-5.3-flash: 1,048,576 in / 131,072 out.

Does the zai API support function calling?

Yes — the zai API supports function calling.

Where does China AI Hub get its zai API data?

From 2 sources, last verified 2026-09-20.