GLM-5.2
CodeFreemium

GLM-5.2

Z.ai's open-weight, MIT-licensed coding model — a 744B-parameter Mixture-of-Experts with a 1M-token context.

4.6
Try it now →

About GLM-5.2

GLM-5.2 (Z.ai, formerly Zhipu AI) is an MIT-licensed Mixture-of-Experts model with ~40B active parameters across 384 experts and a 1-million-token context window. It is currently the strongest open-weight model on agentic coding benchmarks, beating GPT-5.5 on SWE-bench Pro and trailing Claude Opus 4.8 by about a point on FrontierSWE, at roughly one-sixth the cost.

Tags

codingopen-weightmoemit-licensechina1m-context

Reviews (1)

GLM-5.2 Review (June 2026): Open Weights That Out-Code GPT-5.5

4.6

GLM-5.2 is the strongest open-weight coding model yet: MIT-licensed, 744B MoE, 1M context. It beats GPT-5.5 on SWE-bench Pro and trails Opus 4.8 by a point on FrontierSWE — at a fraction of the cost. Our review, with benchmark charts.

Pros:
  • +Strongest open-weight coding model
  • +MIT-licensed and self-hostable
  • +1M-token context fits whole repos
Cons:
  • -Falls behind on agentic Tool-Decathlon
  • -Hosted API routes data through China
  • -No benchmarks published at launch
Read full review →

Similar Tools

Claude Code

Claude Code

Code
4.8

Anthropic's terminal-first agentic coding tool — a CLI agent that reads your repo, runs tests, edits files and opens PRs, built for long autonomous jobs.

PaidTry it →
Cursor

Cursor

Code
4.7

AI-native IDE that transforms the development experience. Write code through conversation.

FreemiumTry it →
OpenAI Codex

OpenAI Codex

Code
4.7

OpenAI's coding agent that shows up everywhere — terminal, IDE, ChatGPT cloud and GitHub — sharing one account and context, with ~4M weekly developers.

PaidTry it →
← Back to All Tools