Benchmark Almanac

MedXpertQA MM Extended

Claude Opus 5.5 71.2% leads since September 2026
Not in model cards

Models answer multimodal medical questions over clinical images and patient details. Mercor adds 51 new expert-written tasks to the 200 public ones and reports both scores.

Held-out extension of medxpertqa. 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
Medicine
Level
Expert
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
42
Entries on its own leaderboard, 41 models we track

Claude Opus 5.5 leads with 71.2%

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.

42 entries. Scroll sideways for the rest.

  1. 71.2 Claude Opus 5.5
  2. 68.0 GPT-6 Sol
  3. 66.7 Gemini 4 Argon
  4. 66.0 Claude Fable 5.1
  5. 65.7 Muse Wonder
  6. 64.7 GPT-5.6 Sol
  7. 64.7 GPT-6.1 Sol
  8. 64.1 Claude Opus 5
  9. 62.7 Gemini 3.7 Flash
  10. 62.1 GPT-5.4
  11. 60.8 Gemini 3.1 Pro
  12. 60.8 GPT-5.5
  13. 60.8 Gemini 3.8 Flash
  14. 60.8 GPT-6 Astra
  15. 60.1 Muse Spark 1.3
  16. 58.2 Muse Spark 1.1
  17. 58.2 Muse Spark 1.2
  18. 57.5 Gemini 3.5 Flash
  19. 56.9 Claude Fable 5
  20. 56.2 GPT-5.6 Luna
  21. 55.9 Muse Spark Open
  22. 55.6 Claude Sonnet 5.5
  23. 54.2 Gemini 3.6 Flash
  24. 52.9 GLM-5.3-Flash
  25. 52.0 Grok 4.7
  26. 51.6 Claude Opus 4.8
  27. 51.6 Kimi K2.7 Code
  28. 51.6 Qwen 3.8 Max
  29. 51.0 Claude Opus 4.7
  30. 51.0 GPT-5.6 Terra
  31. 51.0 Inkling
  32. 48.4 Grok 4.6
  33. 48.4 DeepSeek V4.1 Flash
  34. 48.4 Claude Haiku 5.5
  35. 47.7 Kimi K3
  36. 47.7 Gemini 3.5 Flash-Lite
  37. 46.4 GPT-6 Luna
  38. 45.1 Qwen3.5
  39. 43.8 Claude Sonnet 5
  40. 41.2 Claude Sonnet 4.6
  41. 41.2 Grok 4.5
  42. 34.6 MiniMax-M3
  1. 1 Claude Opus 5.5 max 71.2%
  2. 2 GPT-6 Sol max 68%
  3. 3 Gemini 4 Argon high 66.7%
  4. 4 Claude Fable 5.1 high 66%
  5. 5 Muse Wonder max 65.7%
  6. 6 GPT-5.6 Sol max 64.7%
  7. 7 GPT-6.1 Sol max 64.7%
  8. 8 Claude Opus 5 max 64.1%
  9. 9 Gemini 3.7 Flash high 62.7%
  10. 10 GPT-5.4 xhigh 62.1%
  11. 11 Gemini 3.1 Pro high 60.8%
  12. 12 GPT-5.5 xhigh 60.8%
  13. 13 Gemini 3.8 Flash high 60.8%
  14. 14 GPT-6 Astra max 60.8%
  15. 15 Muse Spark 1.3 max 60.1%
  16. 16 Muse Spark 1.1 xhigh 58.2%
  17. 17 Muse Spark 1.2 xhigh 58.2%
  18. 18 Gemini 3.5 Flash high 57.5%
  19. 19 Claude Fable 5 max 56.9%
  20. 20 GPT-5.6 Luna max 56.2%
  21. 21 Muse Spark Open xhigh 55.9%
  22. 22 Claude Sonnet 5.5 max 55.6%
  23. 23 Gemini 3.6 Flash high 54.2%
  24. 24 GLM-5.3-Flash max 52.9%
  25. 25 Grok 4.7 xhigh 52%
  26. 26 Claude Opus 4.8 max 51.6%
  27. 27 Kimi K2.7 Code high 51.6%
  28. 28 Qwen 3.8 Max xhigh 51.6%
  29. 29 Claude Opus 4.7 max 51%
  30. 30 GPT-5.6 Terra max 51%
  31. 31 Inkling high 51%
  32. 32 Grok 4.6 xhigh 48.4%
  33. 33 DeepSeek V4.1 Flash max 48.4%
  34. 34 Claude Haiku 5.5 max 48.4%
  35. 35 Kimi K3 max 47.7%
  36. 36 Gemini 3.5 Flash-Lite high 47.7%
  37. 37 GPT-6 Luna max 46.4%
  38. 38 Qwen3.5 45.1%
  39. 39 Claude Sonnet 5 max 43.8%
  40. 40 Claude Sonnet 4.6 high 41.2%
  41. 41 Grok 4.5 high 41.2%
  42. 42 MiniMax-M3 high 34.6%

Source: Official MedXpertQA MM Extended leaderboard

#EntryScoreSettingDateSource
1 Claude Opus 5.5 71.2% max Sep 2026 Official leaderboard
2 GPT-6 Sol 68% max Sep 2026 Official leaderboard
3 Gemini 4 Argon 66.7% high Sep 2026 Official leaderboard
4 Claude Fable 5.1 66% high Sep 2026 Official leaderboard
5 Muse Wonder 65.7% max Oct 2026 Official leaderboard
6 GPT-5.6 Sol 64.7% max Jul 2026 Official leaderboard
7 GPT-6.1 Sol 64.7% max Sep 2026 Official leaderboard
8 Claude Opus 5 64.1% max Jul 2026 Official leaderboard
9 Gemini 3.7 Flash 62.7% high Aug 2026 Official leaderboard
10 GPT-5.4 62.1% xhigh Mar 2026 Official leaderboard
11 Gemini 3.1 Pro 60.8% high Feb 2026 Official leaderboard
12 GPT-5.5 60.8% xhigh Apr 2026 Official leaderboard
13 Gemini 3.8 Flash 60.8% high Sep 2026 Official leaderboard
14 GPT-6 Astra 60.8% max Sep 2026 Official leaderboard
15 Muse Spark 1.3 60.1% max Sep 2026 Official leaderboard
16 Muse Spark 1.1 58.2% xhigh Jul 2026 Official leaderboard
17 Muse Spark 1.2 58.2% xhigh Aug 2026 Official leaderboard
18 Gemini 3.5 Flash 57.5% high May 2026 Official leaderboard
19 Claude Fable 5 56.9% max Jun 2026 Official leaderboard
20 GPT-5.6 Luna 56.2% max Jul 2026 Official leaderboard
21 Muse Spark Open 55.9% xhigh Oct 2026 Official leaderboard
22 Claude Sonnet 5.5 55.6% max Sep 2026 Official leaderboard
23 Gemini 3.6 Flash 54.2% high Jul 2026 Official leaderboard
24 GLM-5.3-Flash 52.9% max Aug 2026 Official leaderboard
25 Grok 4.7 52% xhigh Sep 2026 Official leaderboard
26 Claude Opus 4.8 51.6% max May 2026 Official leaderboard
27 Kimi K2.7 Code 51.6% high Jun 2026 Official leaderboard
28 Qwen 3.8 Max 51.6% xhigh Aug 2026 Official leaderboard
29 Claude Opus 4.7 51% max Apr 2026 Official leaderboard
30 GPT-5.6 Terra 51% max Jul 2026 Official leaderboard
31 Inkling 51% high Jul 2026 Official leaderboard
32 Grok 4.6 48.4% xhigh Aug 2026 Official leaderboard
33 DeepSeek V4.1 Flash 48.4% max Sep 2026 Official leaderboard
34 Claude Haiku 5.5 48.4% max Oct 2026 Official leaderboard
35 Kimi K3 47.7% max Jul 2026 Official leaderboard
36 Gemini 3.5 Flash-Lite 47.7% high Jul 2026 Official leaderboard
37 GPT-6 Luna 46.4% max Sep 2026 Official leaderboard
38 Qwen3.5 45.1% Official leaderboard
39 Claude Sonnet 5 43.8% max Jun 2026 Official leaderboard
40 Claude Sonnet 4.6 41.2% high Feb 2026 Official leaderboard
41 Grok 4.5 41.2% high Jul 2026 Official leaderboard
42 MiniMax-M3 34.6% 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 Anthropic
Own leaderboard 3 leaders
Claude Opus 5.5 71.2%

Source: Official MedXpertQA MM 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%) Claude Opus 5.5 Claude Fable 5.1 GPT-5.4 Claude Sonnet 4.6 0% 20% 40% 60% 80% 100% 2026 Near the ceiling (85%) Claude Opus 5.5 GPT-5.6 Sol

Source: Official MedXpertQA MM 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 Law, finance and medicine

All 179 in Law, finance and medicine
  • MedXpertQA-MM In use Models answer expert-level medical exam questions that combine images with clinical records and test results. Scoring is accuracy. Qwen 3.8 Max 80.4%
  • HealthBench Professional In use 525 physician-authored tasks from real clinician chats (care consult, documentation, research) with physician-adjudicated rubrics. Claude Sonnet 5.5 69.2%
  • HealthBench In use 5,000 realistic health conversations graded against physician-written rubrics. A length-adjusted variant penalizes verbosity. GPT-5 67.2%
  • FinSearchComp In use Agents search the web to fetch live financial data, look up history, and run multi-step investigations, across 635 expert-written questions. Scored by answer accuracy. Ling-3.1-flash 81.7%
  • KBL In use Models answer Korean legal knowledge and reasoning tasks and Korean bar exam questions. It has 7 knowledge tasks, 4 reasoning tasks and 2,510 bar exam items. Solar Pro 4 77.5%
  • KorMedMCQA In use Models answer 7,469 multiple-choice questions from Korean licensing exams for doctors, nurses, pharmacists and dentists, held from 2012 to 2024. Solar Pro 4 93.2%