GLM-5.2
Z.AITEXT Open weight
- Provider
- Z.AI
- Modality
- TEXT
- Parameters
- 744B
- Context window
- 1,000,000 tokens
- Weights
- Open
- Released
- 16 Jun 2026
GLM-5.2 (June 2026) is Z.AI's strongest open-weight model for long-horizon autonomous coding — a 744B MoE with a stable 1M-token context, released under MIT.
Best for
- Long-horizon autonomous / agentic coding
- Whole-repo work with a stable 1M-token context
- Self-hostable frontier-class open model (MIT)
- Works in Claude Code, Cline, Kilo Code and more
How it compares — and the India angle
- Beats GPT-5.5 on several long-horizon coding benchmarks at ~1/6th the cost — a serious value pick for Indian dev teams.
- MIT open weights (744B) mean full data control and self-hosting, though the size needs a real GPU cluster.
- New IndexShare architecture cuts 1M-context compute ~2.9x, making long-context cheaper.
Benchmarks
Representative public scores (approximate, higher is better) for relative comparison. Check the provider for the latest official results.
GPQA Diamond91
AIME (math)99
SWE-bench Pro62
Terminal-Bench81
How to access
MIT open weights (744B-A40B). API ~$1.40 / 1M input, ~$4.40 / 1M output. Coding plans from ~$12.60/mo.
Access GLM-5.2