Kimi K3.1 Pro Scores 43.3 on CrusaderBench, Just Below Claude Opus 5.5
Hesamation · x · 2026-10-24
CrusaderBench is an independent benchmark that has models run a medieval village in Stronghold Crusader for 33–45 game minutes per map, seeing the game via screenshots and acting with mouse and keyboard; scores are measured against doing nothing and against a human game on the same map (80% net worth, 10% population, 10% popularity).
- Kimi K3.1 Pro scored 43.3 on The Sinkhole (no arms) and 42.5 on Large Barren Desert (no arms), plus 31.8 on standard The Sinkhole — just below Claude Opus 5.5 and level with Sonnet 5.5.
- All models ran at "medium" reasoning; Kimi was set to "high" since it lacks a medium tier. Tokens per turn on the same map: Kimi 550, Gemini 4 Argon 1,340, Opus 5.5 290, GPT-6.1 Sol 100.
- The site also plots a score-vs-cost Pareto frontier and population growth, comparing only models with full-budget scores on all three maps.
- In post-game interviews, Claude Fable 5.1 admitted it "never checked the actual conversion ratios; I reasoned from unit prices and got it backwards."
Related event: Kimi K3.1 Pro Tested on CrusaderBench, Trails Opus 5.5(2 posts)→
More from Models
- Scale CEO teases new models as users say Muse Wonder Full is underrated — alexandr_wang · 2026-10-24
- Canva Craft trained on millions of templates and 25,000 designer-evaluated image pairs — jacob_top · 2026-10-24
- $34.72 of Kimi K3.1 Pro tokens builds a playable Spider-Man Manhattan in 8 hours — sven_ai · 2026-10-24
- Cisco engineer replicates free Antares security model at ~80% of official score — shashib · 2026-10-24
- Kimi K3.1 vs Opus 5.5: 80% Cheaper and 4x Faster on Same Task — sven_ai · 2026-10-24
- Claim: Open-source Kimi K3.1 Pro beats OpenAI's GPT-6-Astra, unverified — sven_ai · 2026-10-24