GLM-5.1 Tops SWE-bench Pro — First Open-Source Model to Lead a Major Coding Benchmark

NerdLevelTech / freeCodeCamp / Latent Space·high signal

Zhipu AI released GLM-5.1 under the MIT license — a 754B-parameter MoE model with 40B active parameters and 200K context, scoring 58.4% on SWE-bench Pro to edge past GPT-5.4 (57.7%) and Claude Opus 4.6 (57.3%). First open-weight model to ever lead that leaderboard. Chinese labs (Zhipu AI, Alibaba, Moonshot AI, DeepSeek) now hold most top open-weight positions, with Google's Gemma 4 31B breaking into the top 5. Zhipu became the first publicly traded foundation model company after its HKD 4.35B Hong Kong IPO in January 2026.