Zhipu AI's GLM-5.2 trails closed-source leaders on coding tasks

Zhipu AI released GLM-5.2, an open-source model with a stable 1-million-token context under the MIT license. On long-horizon coding benchmarks, it scores 74.4% on FrontierSWE, one point behind Anthropic's Claude Opus 4.8. The model beats GPT-5.5 on PostTrainBench but reaches only half of Opus 4.8's score on SWE-Marathon.