Zhipu AI (GLM) API
Zhipu AI (GLM) API
Zhipu AI API provides GLM series LLMs (GLM-4 general-purpose, GLM-4V multimodal, CodeGeeX code, AutoGLM autonomous agents). Bearer token authentication, chat completions endpoint supports streaming and function calling.
Tsinghua academic heritage (strong in academic scenarios)
General LLM capability slightly weaker than Qwen-Max / GPT-4
Sign up at bigmodel.cn for API key. POST https://open.bigmodel.cn/api/paas/v4/chat/completions with model: "glm-4" + messages.
Uptime · 30-day window
About this API
Zhipu AI is an AI company incubated from Tsinghua University Computer Science KEG Lab, founded 2019. Early open-sourcing of ChatGLM-6B caused a sensation in the Chinese LLM open-source community, accelerating China's LLM open-source ecosystem. Commercialization via Open Platform (bigmodel.cn): GLM-4 (flagship), GLM-4V (multimodal), CodeGeeX (code generation), AutoGLM (GUI agent that auto-operates PC/mobile to execute tasks). Customers mainly academic institutions, government research departments, ToB enterprises. Together with Qwen / Moonshot Kimi / DeepSeek forms China's first-tier LLM, but current commercialization scale smaller than Alibaba Cloud's Qwen.
What you can build
- 1Chinese LLM apps (especially academic / tech scenarios)
- 2Multimodal visual understanding (GLM-4V)
- 3AutoGLM automated PC / mobile operations
- 4Code generation (CodeGeeX)
Strengths & limitations
Strengths
- Tsinghua academic heritage (strong in academic scenarios)
- Open-source ChatGLM series influence (60k+ GitHub stars)
- AutoGLM is China's GUI agent pioneer
- Cheap pricing (GLM-4 input ¥0.1/1000 tokens)
Limitations
- General LLM capability slightly weaker than Qwen-Max / GPT-4
- Ecosystem maturity less than OpenAI
Official quickstart
Read the official quickstart at open.bigmodel.cn.
Getting started
Sign up at bigmodel.cn for API key. POST https://open.bigmodel.cn/api/paas/v4/chat/completions with model: "glm-4" + messages.
FAQ
What is AutoGLM?+
One of China's earliest GUI agents — lets LLM actually "operate" your browser/phone (click, type, navigate), auto-completing multi-step tasks like booking flights / looking up info.
Technical details
- Auth type
- api_key
- Pricing
- paid
- Rate limit
- Default RPM by paid tier; request higher in console.
- Free tier quota
- GLM-4-Flash is permanently free for all users. New accounts also receive ~25M tokens of paid-model credit on signup.
- Protocols
- REST
- SDKs
- python, javascript, java
- Response time
- 1050 ms
- Last health check
- 6/26/2026, 6:22:42 AM