SO-Arena v1.0 - Retrieval & Reranking: leaderboard

SO-Arena v1.0 Retrieval & Reranking 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
1KnowLine-31B-Gen164
2Jeva-4B64
3Sieve-31B-v363.5
4GPT-6 SolMedium62.5#26
5GPT-6 LunaMedium62#99
6Clef 27B61
7H2O-Lightning-4B61
8Jev 1.14.059.5
9Kev 27B59
10Clef flash 9.5B57.5
11JevAny-27B57
12GLiDE v1.257
13PPLX-Decider-Flash-v1.256
14PPLX-Decider-v1.256
15StartLux-v1.1-4B-AutoThink56

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