SO-Arena v1.0 - Scientific Reasoning: leaderboard

SO-Arena v1.0 Scientific Reasoning domain score on the source-selected1232-task panel. Domain scores average native task credits for static decisions, tool outcomes or web progress, then scale to0–100. The full release has2919tasks; this leaderboard emphasizes the highest difficulty tiers and keeps all untiered sentiment/knowledge tasks. Metric: Native panel task credit mean (%). Higher is better.

Metric: Native panel task credit mean (%). Source: huggingface.co. Saturation forecast: Around February 2027. 20 models tracked. Updated 14 Nov 2026.

Top models

#ModelSettingScoreOverall rank
1GPT-6 SolMedium91.25#26
2GPT-6 LunaMedium87.5#99
3Jev 1.14.071.25
4JevAny-27B60
5Jeva-4B60
6GLiDE v1.255
7Sieve-31B-v355
8Kev 27B53.75
9PPLX-Decider-v1.252.5
10KnowLine-31B-Gen152.5
11Clef 27B51.25
12StartLux-v1.1-4B-AutoThink51.25
13Clef flash 9.5B47.5
14StartLux-v1.1-4B-Instant46.25
15SII-Decision-4B45

No result here: #1 Claude Fable 5.2, #2 Gemini 4 Carbon, #5 Claude Opus 5.5.

Interactive version: theaggregate.ai/benchmark?slug=so-arena-v1-0-scientific-reasoning · How It Works · Data refreshed daily, snapshot 2026-11-16.