Chinese lab Z AI just released GLM-5.2, an impressive new open weights model with a 1M token context...
TL;DR · AI Summary
Z AI 发布 GLM-5.2，支持 1M token 上下文窗口，性能超越 GPT-5.5 和 Opus 4.8。
Key Takeaways
- GLM-5.2 在长程编程任务中得分为 74.4，优于 GPT-5.5 的 72.6。
- GLM-5.2 在 SWE-bench 上得分为 62.1，在 AIME 2026 数学测试中得分为 99.2。
- GLM-5.2 使用 MIT 许可证发布，实现“无边界技术访问”。
Outline
Jump quickly between sections.
Mindmap
See how the topics connect at a glance.
查看大纲文本（无障碍 / 无 JS 友好）
- GLM-5.2 发布
- 性能表现
- 长程编程：74.4
- SWE-bench：62.1
- AIME 2026 数学测试：99.2
- 开源与许可
- MIT 许可证
Highlights
Key sentences worth saving and sharing.
GLM-5.2 在长程编程任务中得分为 74.4，优于 GPT-5.5 的 72.6。
GLM-5.2 在 AIME 2026 数学测试中得分为 99.2，超越 Opus 4.8 和 GPT-5.5。
Z AI 使用 MIT 许可证发布 GLM-5.2，实现“无边界技术访问”。
The Rundown AI on X: "Chinese lab Z AI just released GLM-5.2, an impressive new open weights model with a 1M token context window. A few benchmark comparisons, slotting between Opus 4.8 and GPT 5.5 on several fronts: - 74.4 on long-horizon coding, ahead of GPT-5.5's 72.6. - 62.1 on SWE-bench https://t.co/fVIM88NSHG" / X
The Rundown AI
@TheRundownAI
Chinese lab Z AI just released GLM-5.2, an impressive new open weights model with a 1M token context window. A few benchmark comparisons, slotting between Opus 4.8 and GPT 5.5 on several fronts: - 74.4 on long-horizon coding, ahead of GPT-5.5's 72.6. - 62.1 on SWE-bench
gain. - 99.2 on the AIME 2026 math set, ahead of both Opus 4.8 and GPT-5.5. GLM-5.2 also jumped to No. 1 in the
@
Designarena
, overtaking Fable(!).
Zai_org
shipped the weights under an MIT license, with its launch post calling it "technical access without borders."
6:39 PM · Jun 16, 2026
3.1K
Views
1
11
4
3
34
8
Read 11 replies