← All benchmarks

Best AI for reasoning & knowledge (Nov 2026)

Expert exams, graduate science, contest maths and broad knowledge. Pick one of 4 leaderboards; the leaders on each, with sources, are summarised below the chart.

Humanity's Last Exam

2,500 of the toughest, subject-diverse, multi-modal questions, built as a last academic exam for AI.

Scale AI and Center for AI Safety does not publish a cost per test for this benchmark, so there is no performance-vs-cost chart. Showing the ranking instead.

Highest reported score: GPT 6 Astra (OpenAI) at 54.80%.

Accuracy, highest first. Position is the source’s published rank.

  1. 1GPT 6 Astra OpenAI54.80%
  2. 2Fable 5.1 (xhigh) Anthropic46.50%
  3. 3gemini-3.1-pro-preview (thinking high) Google46.44%
  4. 4Sonnet 5.5 (xhigh) Anthropic46.12%
  5. 5Gemini 3.8 Flash Google44.52%
  6. 6gpt-5.4-pro-2026-03-05 OpenAI44.32%
  7. 7Muse Spark Meta40.56%
  8. 8gemini-3-pro-preview Google37.52%
  9. 9gpt-5.4-2026-03-05 (xhigh thinking) OpenAI36.24%
  10. 10claude-opus-4-7 Anthropic36.20%
  11. 11claude-opus-4-6-thinking-max Anthropic34.44%
  12. 12gpt-5-pro-2025-10-06 OpenAI31.64%
  13. 13Claude Haiku 5.5 Anthropic30.12%
  14. 14gpt-5.2-2025-12-11 OpenAI27.80%
  15. 15gpt-5-2025-08-07 OpenAI25.32%
  16. 16claude-opus-4-5-20251101-thinking Anthropic25.20%
  17. 17kimi-k2.5 Moonshot24.37%
  18. 18gpt-5.1-thinking OpenAI23.68%
  19. 19gemini-2.5-pro-preview-06-05 Google21.64%
  20. 20o3 (high) (April 2025) OpenAI20.32%
Humanity's Last Exam: all published values (20 models). “–” means the source lists no value.
1GPT 6 AstraOpenAI54.80%39%
2Fable 5.1 (xhigh)Anthropic46.50%20%
3gemini-3.1-pro-preview (thinking high)Google46.44%51%
4Sonnet 5.5 (xhigh)Anthropic46.12%–
5Gemini 3.8 FlashGoogle44.52%51%
6gpt-5.4-pro-2026-03-05OpenAI44.32%38%
7Muse SparkMeta40.56%50%
8gemini-3-pro-previewGoogle37.52%57%
9gpt-5.4-2026-03-05 (xhigh thinking)OpenAI36.24%42%
10claude-opus-4-7Anthropic36.20%47%
11claude-opus-4-6-thinking-maxAnthropic34.44%46%
12gpt-5-pro-2025-10-06OpenAI31.64%49%
13Claude Haiku 5.5Anthropic30.12%–
14gpt-5.2-2025-12-11OpenAI27.80%45%
15gpt-5-2025-08-07OpenAI25.32%50%
16claude-opus-4-5-20251101-thinkingAnthropic25.20%55%
17kimi-k2.5Moonshot24.37%67%
18gpt-5.1-thinkingOpenAI23.68%55%
19gemini-2.5-pro-preview-06-05Google21.64%72%
20o3 (high) (April 2025)OpenAI20.32%34%

Source: Scale AI and Center for AI Safety · checked 17 Nov 2026

Open the full Humanity's Last Exam leaderboard

Results from published AI leaderboards. Compare models within each benchmark; scores from different tests aren’t directly comparable. How we collect these results

Highest reported scores

As of our last check on 17 Nov 2026 (the source shows no leaderboard date), the highest reported score on Humanity's Last Exam is GPT 6 Astra (OpenAI) at 54.80%, per Scale AI and Center for AI Safety. As of 1 Sep 2026, the highest reported score on GPQA Diamond is Gemini 3.1 Pro Preview (02/26) (Google) at 95.45%, per Vals AI. The source no longer updates this leaderboard. As of 16 Apr 2026, the highest reported score on AIME is Gemini 3.1 Pro Preview (02/26) (Google) at 98.13%, per Vals AI. The source no longer updates this leaderboard. As of 1 Sep 2026, the highest reported score on MMLU Pro is Claude Fable 5.1 (Anthropic) at 92.38%, per Vals AI. The source no longer updates this leaderboard. No single model has the highest reported score on every reasoning & knowledge leaderboard here, so compare within each benchmark.

All 4 reasoning & knowledge benchmarks

Other published views: Accuracy vs calibration
  • Anthropic
  • OpenAI
  • Google
  • Meta
  • Moonshot
Humanity's Last Exam: Calibration error (lower is better) vs AccuracyScatter plot of 18 models. Leader on Calibration error (lower is better): Fable 5.1 (xhigh) at 20%. Full values in the table below.20%30%40%50%20%40%60%ACCURACYCALIBRATION ERROR (LOWER IS BETTER) · FLIPPED, BETTER IS UPo3 (high) (April 2025) (OpenAI): Accuracy 20.32%, Calibration error (lower is better) 34%Ogemini-2.5-pro-preview-06-05 (Google): Accuracy 21.64%, Calibration error (lower is better) 72%Ggpt-5.1-thinking (OpenAI): Accuracy 23.68%, Calibration error (lower is better) 55%Okimi-k2.5 (Moonshot): Accuracy 24.37%, Calibration error (lower is better) 67%Kclaude-opus-4-5-20251101-thinking (Anthropic): Accuracy 25.20%, Calibration error (lower is better) 55%Agpt-5-2025-08-07 (OpenAI): Accuracy 25.32%, Calibration error (lower is better) 50%Ogpt-5.2-2025-12-11 (OpenAI): Accuracy 27.80%, Calibration error (lower is better) 45%Ogpt-5-pro-2025-10-06 (OpenAI): Accuracy 31.64%, Calibration error (lower is better) 49%Oclaude-opus-4-6-thinking-max (Anthropic): Accuracy 34.44%, Calibration error (lower is better) 46%Aclaude-opus-4-7 (Anthropic): Accuracy 36.20%, Calibration error (lower is better) 47%Agpt-5.4-2026-03-05 (xhigh thinking) (OpenAI): Accuracy 36.24%, Calibration error (lower is better) 42%Ogemini-3-pro-preview (Google): Accuracy 37.52%, Calibration error (lower is better) 57%GMuse Spark (Meta): Accuracy 40.56%, Calibration error (lower is better) 50%Mgpt-5.4-pro-2026-03-05 (OpenAI): Accuracy 44.32%, Calibration error (lower is better) 38%OGemini 3.8 Flash (Google): Accuracy 44.52%, Calibration error (lower is better) 51%Ggemini-3.1-pro-preview (thinking high) (Google): Accuracy 46.44%, Calibration error (lower is better) 51%GFable 5.1 (xhigh) (Anthropic): Accuracy 46.50%, Calibration error (lower is better) 20%AGPT 6 Astra (OpenAI): Accuracy 54.80%, Calibration error (lower is better) 39%OFable 5.1GPT 6 Astrao3 (high)
Humanity's Last Exam: Calibration error (lower is better) vs AccuracyScatter plot of 18 models. Leader on Calibration error (lower is better): Fable 5.1 (xhigh) at 20%. Full values in the table below.20%30%40%50%20%40%60%ACCURACYCALIBRATION ERROR (LOWER IS BETTER) · FLIPPED, BETTER IS UPo3 (high) (April 2025) (OpenAI): Accuracy 20.32%, Calibration error (lower is better) 34%Ogemini-2.5-pro-preview-06-05 (Google): Accuracy 21.64%, Calibration error (lower is better) 72%Ggpt-5.1-thinking (OpenAI): Accuracy 23.68%, Calibration error (lower is better) 55%Okimi-k2.5 (Moonshot): Accuracy 24.37%, Calibration error (lower is better) 67%Kclaude-opus-4-5-20251101-thinking (Anthropic): Accuracy 25.20%, Calibration error (lower is better) 55%Agpt-5-2025-08-07 (OpenAI): Accuracy 25.32%, Calibration error (lower is better) 50%Ogpt-5.2-2025-12-11 (OpenAI): Accuracy 27.80%, Calibration error (lower is better) 45%Ogpt-5-pro-2025-10-06 (OpenAI): Accuracy 31.64%, Calibration error (lower is better) 49%Oclaude-opus-4-6-thinking-max (Anthropic): Accuracy 34.44%, Calibration error (lower is better) 46%Aclaude-opus-4-7 (Anthropic): Accuracy 36.20%, Calibration error (lower is better) 47%Agpt-5.4-2026-03-05 (xhigh thinking) (OpenAI): Accuracy 36.24%, Calibration error (lower is better) 42%Ogemini-3-pro-preview (Google): Accuracy 37.52%, Calibration error (lower is better) 57%GMuse Spark (Meta): Accuracy 40.56%, Calibration error (lower is better) 50%Mgpt-5.4-pro-2026-03-05 (OpenAI): Accuracy 44.32%, Calibration error (lower is better) 38%OGemini 3.8 Flash (Google): Accuracy 44.52%, Calibration error (lower is better) 51%Ggemini-3.1-pro-preview (thinking high) (Google): Accuracy 46.44%, Calibration error (lower is better) 51%GFable 5.1 (xhigh) (Anthropic): Accuracy 46.50%, Calibration error (lower is better) 20%AGPT 6 Astra (OpenAI): Accuracy 54.80%, Calibration error (lower is better) 39%OFable 5.1GPT 6 Astrao3 (high)
Humanity's Last Exam: top 15 by AccuracyBars from zero. Leader: GPT 6 Astra at 54.80%. Full values in the table below.0%20%40%60%GPT 6 Astra (OpenAI), rank 1: 54.80%OGPT 6 Astra54.80%Fable 5.1 (xhigh) (Anthropic), rank 2: 46.50%AFable 5.146.50%gemini-3.1-pro-preview (thinking high) (Google), rank 3: 46.44%Ggemini-3.1-pro-preview46.44%Sonnet 5.5 (xhigh) (Anthropic), rank 4: 46.12%ASonnet 5.546.12%Gemini 3.8 Flash (Google), rank 5: 44.52%GGemini 3.8 Flash44.52%gpt-5.4-pro-2026-03-05 (OpenAI), rank 6: 44.32%Ogpt-5.4-pro-2026-03-0544.32%Muse Spark (Meta), rank 7: 40.56%MMuse Spark40.56%gemini-3-pro-preview (Google), rank 8: 37.52%Ggemini-3-pro-preview37.52%gpt-5.4-2026-03-05 (xhigh thinking) (OpenAI), rank 9: 36.24%Ogpt-5.4-2026-03-0536.24%claude-opus-4-7 (Anthropic), rank 10: 36.20%Aclaude-opus-4-736.20%claude-opus-4-6-thinking-max (Anthropic), rank 11: 34.44%Aclaude-opus-4-6-thinking-max34.44%gpt-5-pro-2025-10-06 (OpenAI), rank 12: 31.64%Ogpt-5-pro-2025-10-0631.64%Claude Haiku 5.5 (Anthropic), rank 13: 30.12%AClaude Haiku 5.530.12%gpt-5.2-2025-12-11 (OpenAI), rank 14: 27.80%Ogpt-5.2-2025-12-1127.80%gpt-5-2025-08-07 (OpenAI), rank 15: 25.32%Ogpt-5-2025-08-0725.32%
Humanity's Last Exam: top 12 by AccuracyBars from zero. Leader: GPT 6 Astra at 54.80%. Full values in the table below.0%20%40%60%GPT 6 Astra (OpenAI), rank 1: 54.80%OGPT 6 Astra54.80%Fable 5.1 (xhigh) (Anthropic), rank 2: 46.50%AFable 5.146.50%gemini-3.1-pro-preview (thinking high) (Google), rank 3: 46.44%Ggemini-3.1-pro-p…46.44%Sonnet 5.5 (xhigh) (Anthropic), rank 4: 46.12%ASonnet 5.546.12%Gemini 3.8 Flash (Google), rank 5: 44.52%GGemini 3.8 Flash44.52%gpt-5.4-pro-2026-03-05 (OpenAI), rank 6: 44.32%Ogpt-5.4-pro-2026…44.32%Muse Spark (Meta), rank 7: 40.56%MMuse Spark40.56%gemini-3-pro-preview (Google), rank 8: 37.52%Ggemini-3-pro-pre…37.52%gpt-5.4-2026-03-05 (xhigh thinking) (OpenAI), rank 9: 36.24%Ogpt-5.4-2026-03-…36.24%claude-opus-4-7 (Anthropic), rank 10: 36.20%Aclaude-opus-4-736.20%claude-opus-4-6-thinking-max (Anthropic), rank 11: 34.44%Aclaude-opus-4-6-…34.44%gpt-5-pro-2025-10-06 (OpenAI), rank 12: 31.64%Ogpt-5-pro-2025-1…31.64%

Questions people ask

Which AI has the highest reported scores for reasoning & knowledge?

As of our last check on 17 Nov 2026 (the source shows no leaderboard date), the highest reported score on Humanity's Last Exam is GPT 6 Astra (OpenAI) at 54.80%, per Scale AI and Center for AI Safety. As of 1 Sep 2026, the highest reported score on GPQA Diamond is Gemini 3.1 Pro Preview (02/26) (Google) at 95.45%, per Vals AI. The source no longer updates this leaderboard. As of 16 Apr 2026, the highest reported score on AIME is Gemini 3.1 Pro Preview (02/26) (Google) at 98.13%, per Vals AI. The source no longer updates this leaderboard. As of 1 Sep 2026, the highest reported score on MMLU Pro is Claude Fable 5.1 (Anthropic) at 92.38%, per Vals AI. The source no longer updates this leaderboard. No single model has the highest reported score on every reasoning & knowledge leaderboard here, so compare within each benchmark.

Which benchmarks are on this page?

Humanity's Last Exam (run by Scale AI and Center for AI Safety): 2,500 of the toughest, subject-diverse, multi-modal questions, built as a last academic exam for AI. GPQA Diamond (run by Vals AI): Graduate-level “Google-proof” science questions that require deep reasoning. AIME (run by Vals AI): A challenging national maths exam given to top high-school students. MMLU Pro (run by Vals AI): Academic multiple-choice questions covering 14 subjects, including STEM, humanities and social sciences.

How recent are these results?

Humanity's Last Exam: the source shows no leaderboard date; we last checked it on 17 Nov 2026. GPQA Diamond: leaderboard dated 1 Sep 2026, no longer updated by the source; we last checked it on 17 Nov 2026. AIME: leaderboard dated 16 Apr 2026, no longer updated by the source; we last checked it on 17 Nov 2026. MMLU Pro: leaderboard dated 1 Sep 2026, no longer updated by the source; we last checked it on 17 Nov 2026.

Does ChatJudge run these tests?

No. Every score is copied as published by the leaderboard that ran it, with a link to the source. Scores from different benchmarks are not directly comparable.