GLM-5.2 nearly doubles Kimi K3 on a long-horizon browser benchmark
zainhas · x · 2026-07-24
A composite benchmark for long-horizon browser work shows GLM-5.2 well ahead of Kimi K3.
- The chart reports GLM-5.2 at 63% versus Kimi K3 at 31% on Composite-Bench.
- The benchmark targets long-horizon browser tasks with exact-match grading and pinned max reasoning effort.
- Other models in the chart include Claude Fable 5 (74%), Claude Opus 4.8 (73%), Gemini 3.5 Flash (27%), MiniMax-M3 (26%), GPT-5.6 Sol (18%), and Grok-4.5 (9%).
More from Models
- ICML 2026 position paper: LLM 'thinking traces' are a harmful anthropomorphism — rao2z · 2026-10-27
- One stray 'u' and Claude starts spelling like a Brit — cantrell · 2026-10-27
- Cohere's Invent-A-Dataset paper: zero-seed synthetic data beats frontier APIs on quality and diversity — idanbeck · 2026-10-27
- ARC Prize redesigns leaderboard: highest reasoning effort by default, CSV export — mhmazur · 2026-10-27
- Cohere: open-weight Qwen3.8-Flash-Next tried uploading probe file to Beijing cloud bucket in sandbox — wstone_bd · 2026-10-27
- Model refuses wrong psychiatric advice in deliberately misleading test — FellMentKE · 2026-10-27