Kimi K3.1 Pro Tech Blog - OSWorld V2.1: leaderboard

Moonshot KimiK3.1Pro release evaluations; explicit max header, Kimi temperature1/top-p0.95 except specified overrides. KimiAgent harness, official OSWorldv2 repository. Astra75.8 copied from MuseWonder report. Score (%). Copied third-party comparator results are not re-ingested; source-specific harness retained.

Metric: Score (%). Source: www.kimi.com. Saturation forecast: Around June 2027. 4 models tracked.

Top models

#ModelScoreOverall rank
1Claude Opus 5.5 (Max)81.8#4 (Claude Opus 5.5)
2Claude Fable 5.1 (Max)80.7#9 (Claude Fable 5.1)
3DeepSeek V4.1 Flash (1017) (Max)69.3#25 (DeepSeek V4.1 Flash (1017))

No result here: #6 GPT-6 Astra, #7 GPT-6.1 Sol, #8 Gemini 4 Argon.

Interactive version: theaggregate.ai/benchmark?slug=kimi-k3-1-pro-tech-blog-osworld-v2-1 · How It Works · Data refreshed daily, snapshot 2026-10-24.