SO-Arena v1.0 - Math: leaderboard

SO-Arena v1.0 Math 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 SolMedium100#26
2GPT-6 LunaMedium99#99
3StartLux-v1.1-4B-AutoThink71
4KnowLine-31B-Gen162
5Sieve-31B-v360
6GLiDE v1.254
7Jev 1.14.053
8PPLX-Decider-v1.253
9Kev 27B48
10JevAny-27B45
11H2O-Lightning-4B44
12Jeva-4B44
13StartLux-v1.1-4B-Instant42
14Clef flash 9.5B41
15Clef 27B38

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-math · How It Works · Data refreshed daily, snapshot 2026-11-16.