Best for reasoning
GPT-6 Astra leads · 6 fresh boards · counts toward the leader of leaders · Hard questions with checkable answers: science, math, puzzles, expert exams.
12 of 12 models
| # | Model | Score#1s | ||
|---|---|---|---|---|
| 1 |  GPT-6 AstraOpenAItop 10 on 6, #1 on 4 | 91.7 | 91.74 | 6 |
| 2 |  Claude Fable 5.1Anthropictop 10 on 5, #1 on 1 | 71.7 | 71.71 | 5 |
| 3 | Claude Opus 5.5Anthropictop 10 on 3, #1 on 1 | 46.7 | 46.71 | 3 |
| 4 | Claude Fable 5Anthropictop 10 on 3 | 35.0 | 35.00 | 3 |
| 5 | Claude Opus 5Anthropictop 10 on 4 | 28.3 | 28.30 | 4 |
| 5 | GPT-5.6 SolOpenAItop 10 on 3 | 28.3 | 28.30 | 3 |
| 7 |  Gemini 3.8 FlashGoogletop 10 on 2 | 26.7 | 26.70 | 2 |
| 8 | GPT-5.4 ProOpenAItop 10 on 3 | 25.0 | 25.00 | 3 |
| 9 | Gemini 3.1 Pro PreviewGoogletop 10 on 2 | 23.3 | 23.30 | 2 |
| 10 |  Muse Spark 1.3Metatop 10 on 2 | 21.7 | 21.70 | 2 |
| 11 | Gemini 3.7 FlashGoogletop 10 on 2 | 18.3 | 18.30 | 2 |
| 12 | GPT-5.5OpenAItop 10 on 3 | 16.7 | 16.70 | 3 |
Boards
| Board | Score | ||
|---|---|---|---|
 AA IntelligenceArtificial Analysis#1 Claude Opus 5.5 | Claude Opus 5.5 | 57.6 | Sep 24, 2026 |
 HLEScale AI and CAIS#1 GPT-6 Astra | GPT-6 Astra | 54.8% | Sep 24, 2026 |
 GPQA DiamondEpoch AI#1 GPT-6 Astra | GPT-6 Astra | 95.8% | Sep 2, 2026 |
| Epoch ECIEpoch AI#1 GPT-6 Astra | GPT-6 Astra | 166.6 | Sep 24, 2026 |
 ARC-AGI-2ARC Prize#1 GPT-6 Astra | GPT-6 Astra | 95.0% | Sep 22, 2026 |
 LiveBenchLiveBench#1 Claude Fable 5.1 | Claude Fable 5.1 | 83.4% | Jun 25, 2026 |
 MMLU-ProTIGER-Lab#1 Gemini 3.1 Pro Preview | Gemini 3.1 Pro Preview | 91.2%stale | Mar 11, 2026 |
 Open LLM LBHugging FaceRetired | Retired 2025-03-20 | Retired |
The wall
#1 with the vendor’s logo23 podium7 top 10 stale board, not countedArrow keys move, Enter opens
Rank over time
Scores as published by each board on the capture date. Model names and logos belong to their owners; logos via logo.dev.