Z.ai Releases GLM-5.2: 753B MoE Text-Only Open Weights LLM with 1M Context Tops Benchmarks
English summary
Chinese AI lab Z.ai released GLM-5.2 under the MIT license on June 16, 2026, following a subscriber-only launch on June 13. The model is a text-input-only Mixture of Experts LLM with 753 billion total parameters (40 active parameters), a 1.5TB file size, and a 1-million-token context window, up from GLM-5.1’s 200K. On the Artificial Analysis Intelligence Index v4.1, it leads all open weights models with a score of 51, ahead of MiniMax-M3 (44), DeepSeek V4 Pro (44), and Kimi K2.6 (43). It ranks 2nd on the Code Arena WebDev leaderboard for front-end coding tasks, despite lacking image input capabilities. However, Artificial Analysis noted it generates many output tokens per task (43K on average), more than its competitors. The model is accessible via OpenRouter from multiple providers at $1.40 per million input tokens and $4.40 per million output tokens. In informal SVG generation tests, GLM-5.2 produced impressive animated vector illustrations for some prompts but performed worse than GLM-5.1 on others, failing to apply animations in one case.
Chinese summary
中国AI实验室智谱AI于2026年6月16日以MIT许可证开源了GLM-5.2模型，此前于6月13日向编码计划订阅者发布。该模型是纯文本输入的混合专家（MoE）大语言模型，总参数7530亿（活跃参数40个），文件大小1.5TB，上下文窗口达100万token，远超GLM-5.1的20万。在人工分析公司的智力指数v4.1上，它在开源权重模型中位居榜首（得分51），超过MiniMax-M3（44）、DeepSeek V4 Pro（44）和Kimi K2.6（43）。在Code Arena WebDev排行榜上排名第二，专注于前端开发任务，尽管该模型不支持图像输入。但人工分析发现其每个任务平均生成4.3万输出token，高于竞品。用户可通过OpenRouter以每百万输入token 1.40美元、输出token 4.40美元的价格使用该模型。在非正式的SVG生成测试中，GLM-5.2对部分提示词生成了令人印象深刻的动画矢量图，但在其他测试中（如北美负鼠）表现不如GLM-5.1，未能添加动画效果。
Key points
Z.ai released GLM-5.2 under MIT license, a text-only 753B MoE model with 40 active parameters and a 1 million token context window.
智谱AI以MIT许可证开源GLM-5.2，纯文本模型，总参数7530亿（活跃40个），支持100万token上下文。
GLM-5.2 leads the Artificial Analysis Intelligence Index for open weights models, scoring 51, and ranks 2nd on Code Arena WebDev leaderboard.
GLM-5.2在人工分析智力指数中位列开源模型榜首（51分），并在Code Arena WebDev排行榜排名第二。
The model generates excessive output tokens (43K per task on average), more than comparable models like DeepSeek V4 Pro or Kimi K2.6.
模型输出token较多，每个任务平均4.3万，超过DeepSeek V4 Pro和Kimi K2.6等竞品。
Available via OpenRouter at $1.40/M input and $4.40/M output tokens, it is cost-effective compared to proprietary alternatives like GPT-5.5.
通过OpenRouter提供，输入价格每百万token 1.40美元、输出4.40美元，相比GPT-5.5等商用模型成本更低。
Simon Willison's tests show strong SVG animation capabilities for some prompts, but quality regressed on others compared to GLM-5.1.
用户Simon Willison测试表明，模型在部分SVG动画生成上表现优异，但在另一些提示词上效果不如GLM-5.1。