Every AI leaderboard

Leaderboards28Models ranked986

28 leaderboards · 986 models ranked

ARC-AGI-2 leaderboard: who is #1

ARC Prize · Score, higher is better · 92 models · published Nov 6, 2026

ARC Prize

Top 25 of 92 models, best variant per model as published

#ModelScoreCorrect
1GPT-6 Astraas “GPT-6 Astra (Max)”95.0%
2GPT-6.1 Solas “GPT-6.1 Sol (Max)”94.2%
3Claude Opus 5.5as “Claude Opus 5.5 (High)”93.3%
4GPT-5.6 Solas “GPT-5.6 Sol (Max)”92.5%
5New, entered at #5 since Oct 16, 2026Gemini 4 Argonas “Gemini 4 Argon (High)”91.3%
61down from #5 since Oct 16, 2026Claude Opus 5as “Claude Opus 5 (Max)”90.4%
71down from #6 since Oct 16, 2026Claude Fable 5.1as “Claude Fable 5.1 (Max)”90.0%
81down from #7 since Oct 16, 2026GPT-6 Solas “GPT-6 Sol (Max)”89.6%
91down from #8 since Oct 16, 2026Claude Fable 5as “Claude Fable 5 (Max)”89.2%
91down from #8 since Oct 16, 2026Gemini 3.8 Flashas “Gemini 3.8 Flash (High)”89.2%
111down from #10 since Oct 16, 2026GPT-5.5as “GPT-5.5 (XHigh)”85.0%
121down from #11 since Oct 16, 2026Gemini 3.7 Flashas “Gemini 3.7 Flash (High)”84.6%
131down from #12 since Oct 16, 2026Gemini 3as “Gemini 3 Deep Think (2/26)”84.6%
131down from #12 since Oct 16, 2026GPT-5.5 Proas “GPT-5.5 Pro (High)”84.6%
151down from #14 since Oct 16, 2026GPT-5.6 Terraas “GPT-5.6 Terra (Max)”83.9%
16New, entered at #16 since Oct 22, 2026Claude Sonnet 5.5as “Claude Sonnet 5.5 (XHigh)”83.9%
171down from #16 since Oct 22, 2026GPT-5.4 Proas “GPT-5.4 Pro (XHigh)”83.3%
18New, entered at #18 since Nov 7, 2026Inkling 2 Smallas “Inkling 2 Small (XHigh)”81.7%
191down from #18 since Nov 7, 2026DeepSeek V4.1 Flashas “DeepSeek V4.1 Flash 1017 (Max)”78.1%
201down from #19 since Nov 7, 2026Gemini 3.1 Pro Previewas “Gemini 3.1 Pro (Preview)”77.1%
211down from #20 since Nov 7, 2026Dots3 Noteas “Dots3-Note Preview (Max)”76.8%
221down from #21 since Nov 7, 2026Claude 4.7as “Claude 4.7 (Max)”75.8%
231down from #22 since Nov 7, 2026GPT-5.4as “GPT-5.4 (XHigh)”74.0%
241down from #23 since Nov 7, 2026Claude Opus 4.8as “Claude Opus 4.8 (High)”72.1%
241down from #23 since Nov 7, 2026Gemini 3.5 Flashas “Gemini 3.5 Flash (High)”72.1%
About this board and its history

What it measures

Abstract visual puzzles that are easy for people and hard for AI. Semi-private set, systems that are plain models only.

Rank at each month's end

12345678910SepOctNow1GPT-6 Astra2GPT-6.1 Sol3Claude Opus 5.54GPT-5.6 Sol5Gemini 4 Argon6Claude Opus 5
GPT-6 AstraGPT-6.1 SolClaude Opus 5.5GPT-5.6 SolGemini 4 ArgonClaude Opus 5

Other reasoning boards

Source

Scores as published by ARC Prize on the capture date (2026-11-09). Source: arcprize.org leaderboard data files.

arcprize.org/leaderboard