Curiosity · AI Model

GLM-4 Plus

GLM-4 Plus is the flagship model from Zhipu AI, one of China's 'AI Tigers' spun out of Tsinghua University. It offers 128k context, strong bilingual (Chinese/English) performance, and a tool-use stack tuned for agentic workflows. GLM-4 Plus is a default choice for products that need first-class Chinese-language quality.

Model specs

Vendor
Zhipu AI
Family
GLM-4
Released
2024-08
Context window
128,000 tokens
Modalities
text, code
Input price
$7/M tok
Output price
$7/M tok
Pricing as of
2026-04-20

Strengths

  • Top-tier Chinese-language performance on C-Eval and CMMLU
  • Mature tool-use and function-calling interfaces
  • Hosted on the BigModel platform with generous Chinese-market pricing

Limitations

  • English scores slightly below GPT-4-class Western models
  • Access from outside China can be rate-limited
  • Less open tooling than Llama or Qwen ecosystems

Use cases

  • Chinese-language chatbots and customer support
  • Bilingual search and summarisation
  • Domestic (mainland China) enterprise assistants
  • Tool-using agents for e-commerce and CRM

Benchmarks

BenchmarkScoreAs of
C-Eval~83%2026-04
MMLU~82%2026-04
GSM8K~93%2026-04

Frequently asked questions

What is GLM-4 Plus?

GLM-4 Plus is Zhipu AI's flagship proprietary LLM — a bilingual Chinese/English model with 128k context and native tool use, commonly used in Chinese enterprise AI products.

Is GLM-4 Plus open-source?

No. GLM-4 Plus is a closed, hosted model on Zhipu's BigModel API. Zhipu does release smaller open-weight variants (ChatGLM3-6B, GLM-4-9B) on HuggingFace.

Sources

  1. Zhipu AI — GLM-4 docs — accessed 2026-04-20
  2. Zhipu AI on HuggingFace — accessed 2026-04-20