SO-Arena v1.0 - Tool & API Use: leaderboard

SO-Arena v1.0 Tool & API Use 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 LunaMedium68.27#99
2PPLX-Decider-v1.267.31
3GPT-6 SolMedium66.35#26
4GLiDE v1.265.38
5Kev 27B61.54
6Sieve-31B-v359.62
7Jev 1.14.058.65
8JevAny-27B57.69
9KnowLine-31B-Gen155.77
10Clef 27B54.81
11StartLux-v1.1-4B-Instant53.85
12StartLux-v1.1-4B-AutoThink46.15
13PPLX-Decider-Flash-v1.241.35
14JevAny-4B41.35
15H2O-Lightning-4B37.5

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-tool-api-use · How It Works · Data refreshed daily, snapshot 2026-11-16.