GLM-5.1

Trends

• Open-weight models diversify in architecture, modality, and size

Items

1. Apr 7

   Z.ai's GLM-5.1

   A 744B-parameter MIT-licensed MoE (40B active) that tops SWE-Bench Pro at 58.4%, ahead of GPT-5.4, Claude Opus 4.6, and Gemini 3.1 Pro, sustains eight-hour autonomous coding loops, and was trained entirely on Huawei Ascend hardware.