Benchmark Almanac

MMMU-Pro Extended

GPT-6.1 Sol 61.3% leads since September 2026
Not in model cards

Models answer multimodal questions that need both the image and the text. Mercor adds 140 new expert-written tasks to the 1730 public ones and reports both scores.

Held-out extension of mmmu-pro. Mercor shows the original and extended score for each model. The month is approximate. The Wayback Machine first saved the earliest extended pages on August 15th, 2026.

Made by
Mercor
Released
August 2026
Kind
General visual questions
Level
University
Scored
Checked automatically
Answers
Multiple choice
Tasks
Public, plus a private test set
Strongest model tested
GPT-6 Astra #3 on the capability index
Last model added
October 2026
Scores refreshed
November 12th, 2026
39
Entries on its own leaderboard, 38 models we track

GPT-6.1 Sol leads with 61.3%

Best score per entry from the benchmark's own leaderboard, read on November 12th, 2026. Where a model ran with several settings or harnesses, we keep its best. Names in gray aren't in our model list: fine-tunes, agent systems, products and older models. They don't count toward leaders.

39 entries. Scroll sideways for the rest.

  1. 61.3 GPT-6.1 Sol
  2. 60.0 GPT-6 Astra
  3. 51.9 GPT-5.6 Sol
  4. 51.9 GPT-6 Sol
  5. 51.0 Claude Opus 5.5
  6. 47.6 Claude Fable 5.1
  7. 47.4 GPT-5.6 Terra
  8. 44.4 Claude Sonnet 5.5
  9. 43.8 GPT-5.5
  10. 43.2 Muse Wonder
  11. 39.8 GPT-5.4
  12. 39.2 GPT-5.6 Luna
  13. 37.6 Claude Fable 5
  14. 36.4 Gemini 3.7 Flash
  15. 36.4 Gemini 3.8 Flash
  16. 36.0 Kimi K3
  17. 35.5 Muse Spark 1.1
  18. 34.8 Qwen 3.8 Max
  19. 34.8 GPT-6 Luna
  20. 34.3 Claude Opus 5
  21. 33.8 Muse Spark 1.3
  22. 32.5 Muse Spark Open
  23. 31.9 Claude Haiku 5.5
  24. 31.4 Claude Opus 4.8
  25. 31.0 Gemini 3.1 Pro
  26. 30.2 Muse Spark 1.2
  27. 29.8 Grok 4.5
  28. 29.0 Gemini 3.6 Flash
  29. 28.8 Gemini 3.5 Flash
  30. 28.3 Claude Opus 4.7
  31. 26.0 Grok 4.6
  32. 25.0 DeepSeek V4.1 Flash
  33. 24.3 Qwen3.5
  34. 24.0 Inkling
  35. 23.1 Kimi K2.7 Code
  36. 22.6 GLM-5.3-Flash
  37. 20.7 Claude Sonnet 5
  38. 19.9 Claude Sonnet 4.6
  39. 18.1 MiniMax-M3
  1. 1 GPT-6.1 Sol max 61.3%
  2. 2 GPT-6 Astra max 60%
  3. 3 GPT-5.6 Sol max 51.9%
  4. 4 GPT-6 Sol max 51.9%
  5. 5 Claude Opus 5.5 max 51%
  6. 6 Claude Fable 5.1 max 47.6%
  7. 7 GPT-5.6 Terra max 47.4%
  8. 8 Claude Sonnet 5.5 max 44.4%
  9. 9 GPT-5.5 xhigh 43.8%
  10. 10 Muse Wonder max 43.2%
  11. 11 GPT-5.4 xhigh 39.8%
  12. 12 GPT-5.6 Luna max 39.2%
  13. 13 Claude Fable 5 max 37.6%
  14. 14 Gemini 3.7 Flash high 36.4%
  15. 15 Gemini 3.8 Flash high 36.4%
  16. 16 Kimi K3 max 36%
  17. 17 Muse Spark 1.1 xhigh 35.5%
  18. 18 Qwen 3.8 Max xhigh 34.8%
  19. 19 GPT-6 Luna max 34.8%
  20. 20 Claude Opus 5 high 34.3%
  21. 21 Muse Spark 1.3 xhigh 33.8%
  22. 22 Muse Spark Open xhigh 32.5%
  23. 23 Claude Haiku 5.5 max 31.9%
  24. 24 Claude Opus 4.8 high 31.4%
  25. 25 Gemini 3.1 Pro high 31%
  26. 26 Muse Spark 1.2 xhigh 30.2%
  27. 27 Grok 4.5 high 29.8%
  28. 28 Gemini 3.6 Flash high 29%
  29. 29 Gemini 3.5 Flash high 28.8%
  30. 30 Claude Opus 4.7 high 28.3%
  31. 31 Grok 4.6 xhigh 26%
  32. 32 DeepSeek V4.1 Flash max 25%
  33. 33 Qwen3.5 24.3%
  34. 34 Inkling high 24%
  35. 35 Kimi K2.7 Code high 23.1%
  36. 36 GLM-5.3-Flash max 22.6%
  37. 37 Claude Sonnet 5 high 20.7%
  38. 38 Claude Sonnet 4.6 high 19.9%
  39. 39 MiniMax-M3 high 18.1%

Source: Official MMMU-Pro Extended leaderboard

#EntryScoreSettingDateSource
1 GPT-6.1 Sol 61.3% max Sep 2026 Official leaderboard
2 GPT-6 Astra 60% max Sep 2026 Official leaderboard
3 GPT-5.6 Sol 51.9% max Jul 2026 Official leaderboard
4 GPT-6 Sol 51.9% max Sep 2026 Official leaderboard
5 Claude Opus 5.5 51% max Sep 2026 Official leaderboard
6 Claude Fable 5.1 47.6% max Sep 2026 Official leaderboard
7 GPT-5.6 Terra 47.4% max Jul 2026 Official leaderboard
8 Claude Sonnet 5.5 44.4% max Sep 2026 Official leaderboard
9 GPT-5.5 43.8% xhigh Apr 2026 Official leaderboard
10 Muse Wonder 43.2% max Oct 2026 Official leaderboard
11 GPT-5.4 39.8% xhigh Mar 2026 Official leaderboard
12 GPT-5.6 Luna 39.2% max Jul 2026 Official leaderboard
13 Claude Fable 5 37.6% max Jun 2026 Official leaderboard
14 Gemini 3.7 Flash 36.4% high Aug 2026 Official leaderboard
15 Gemini 3.8 Flash 36.4% high Sep 2026 Official leaderboard
16 Kimi K3 36% max Jul 2026 Official leaderboard
17 Muse Spark 1.1 35.5% xhigh Jul 2026 Official leaderboard
18 Qwen 3.8 Max 34.8% xhigh Aug 2026 Official leaderboard
19 GPT-6 Luna 34.8% max Sep 2026 Official leaderboard
20 Claude Opus 5 34.3% high Jul 2026 Official leaderboard
21 Muse Spark 1.3 33.8% xhigh Sep 2026 Official leaderboard
22 Muse Spark Open 32.5% xhigh Oct 2026 Official leaderboard
23 Claude Haiku 5.5 31.9% max Oct 2026 Official leaderboard
24 Claude Opus 4.8 31.4% high May 2026 Official leaderboard
25 Gemini 3.1 Pro 31% high Feb 2026 Official leaderboard
26 Muse Spark 1.2 30.2% xhigh Aug 2026 Official leaderboard
27 Grok 4.5 29.8% high Jul 2026 Official leaderboard
28 Gemini 3.6 Flash 29% high Jul 2026 Official leaderboard
29 Gemini 3.5 Flash 28.8% high May 2026 Official leaderboard
30 Claude Opus 4.7 28.3% high Apr 2026 Official leaderboard
31 Grok 4.6 26% xhigh Aug 2026 Official leaderboard
32 DeepSeek V4.1 Flash 25% max Sep 2026 Official leaderboard
33 Qwen3.5 24.3% Official leaderboard
34 Inkling 24% high Jul 2026 Official leaderboard
35 Kimi K2.7 Code 23.1% high Jun 2026 Official leaderboard
36 GLM-5.3-Flash 22.6% max Aug 2026 Official leaderboard
37 Claude Sonnet 5 20.7% high Jun 2026 Official leaderboard
38 Claude Sonnet 4.6 19.9% high Feb 2026 Official leaderboard
39 MiniMax-M3 18.1% high Jun 2026 Official leaderboard

3 models have held number one on the official leaderboard since its release in August 2026

Lab claims count from the day the card came out, independent runs from the model’s release.

OpenAI
Own leaderboard 3 leaders
GPT-6.1 Sol 61.3%

Source: Official MMMU-Pro Extended leaderboard

Scores over time

All the scores we have, one dot each, so a model tested at several settings appears more than once.

On the benchmark’s own leaderboard Best so far
0% 20% 40% 60% 80% 100% 2026 Near the ceiling (85%) GPT-6.1 Sol GPT-5.6 Sol GPT-5.5 Gemini 3.1 Pro Claude Sonnet 4.6 0% 20% 40% 60% 80% 100% 2026 Near the ceiling (85%) GPT-6.1 Sol GPT-5.6 Sol

Source: Official MMMU-Pro Extended leaderboard

No lab model card reports it yet

We read the headline results of 522 launch posts from 52 labs, and none of them list this benchmark.

More in Vision

All 281 in Vision
  • MMMU In use 11.5K college-level multimodal questions across 30 subjects. Multiple-choice and open accuracy. GPT-5.1 85.4%
  • MMMU-Pro In use Harder MMMU: college-level multimodal questions with more options and vision-only inputs to remove text shortcuts. GPT-5.6 Sol 84.6%
  • CharXiv In use Chart understanding on real arXiv figures. Reasoning questions require synthesizing information from complex charts. Claude Fable 5 93.5%
  • MATH-Vision (MathVision) In use 3,040 competition math problems with visual context. Answer accuracy. Qwen 3.8 Max 97.7%
  • MMStar In use Models answer 1,500 human-curated image questions that cannot be solved without the image, across 6 core capabilities. Scored by accuracy. Qwen 3.8 Max 85.9%
  • ChartQA In use Models answer human-written and generated questions about chart images that need visual and arithmetic reasoning. Scored by relaxed accuracy with a 5 percent numeric tolerance. Command A Vision 90.9%