Z.ai releases GLM-5.3, claims top open-weight coding model

Z.ai released GLM-5.3, a 743-billion-parameter coding model, claiming it's the most capable open-weights coder. It scores 34.5% on Z.ai Code Bench, beating GLM-5.2's 23.4%, but trails closed models like Claude Fable 5 (39.5%).
1 source
Daily brief
Get tomorrow's AI brief in your inbox
More stories today
- Claude Code silently deletes local history older than 30 days
- Split sigmas speed up video reference in Stable Diffusion
- MiniMax tool turns anything into realistic human video
- MirroS' Code-as-World rewrites videos into executable MuJoCo programs
- PRAXIST research system boosts MLE-bench scores 44%