Kimi K3.1 early benchmark leak: WeirdML ~20, blogger predicts benchmaxxed 'soft' model below Sonnet 5.5
teortaxesTex · x · 2026-10-23
Blogger teortaxesTex compiles early scores for Kimi K3.1: WeirdML V3 official ≈20 (final best 28), ECI 164. He hasn't tested it yet but predicts from vibes and priors: 52 AA, 65 Vals, 76 CritPt V2, 49 TB 4.0.
His overall read: the model looks benchmaxxed and "soft" — GDPEval, Harvey, and general ability a bit below Sonnet 5.5; in practice below 6.1 Sol and more expensive per task, though much closer than the 1017 version. Unofficial figures and personal predictions, unverified.
Related event: KOL Predicts Kimi K3.1 Benchmarks Before Testing, Suspects Benchmark Gaming(2 posts)→
More from Models
- Stripping cyber data to handicap models hurts general coding too, warns Lentils80 citing K3.1 Pro — lyraxana · 2026-10-24
- OpenAI researcher slams capacity errors that kill overnight agent sessions with no fallback info — willdepue · 2026-10-24
- Kimi K3.1 Pro debuts: beats Claude Opus 5.5 on Terminal-Bench 4.0 at ~1/3 the cost — Yuchenj_UW · 2026-10-24
- Reddit user on Kimi K3: pinnacle of intelligence, but Qwen wins for daily use — SrijSriv211 · 2026-10-24
- Nathan Lambert: Kimi K3.1 Pro shows Moonshot cracked RL, China keeps pace on capability — natolambert · 2026-10-24
- Omni cuts internal token spend 80% after switching workhorse model to DeepSeek Flash — josh_wills · 2026-10-24