Benchmark Almanac

KorMedMCQA-Mixed

Gemini 3 Pro 97.3% leads since February 2026
Not in model cards

Models take combined text and image exams. The 2022 set contains 281 questions and the 2023 set contains 307. Each year has a separate score.

The authors report total accuracy for 51 models in each year. The inputs combine matching years from KorMedMCQA and KorMedMCQA-V.

Released
February 2026
Kind
Medicine
Level
Expert
Scored
Checked automatically
Answers
Multiple choice
Tasks
All public
Measures
Capability
Language
Korean
In and out
Text, Images → Text
Interaction
Single turn
Refresh
One fixed set
Run by
Anyone
Occupations
Healthcare practitioners and technical
Strongest model tested
GPT-5.2 #51 on the capability index
Last model added
December 2025
Scores refreshed
November 20th, 2026
51
Entries on its own leaderboard, 33 models we track

Gemini 3 Pro leads with 97.3% among models released by December 2025 Gemini 3 Pro leads with 97.3% among models released by December 2025

Best score per entry from the benchmark's own leaderboard, read on November 20th, 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. This view shows only models we track, each at its best.

The newest model with a score here came out in December 2025, and nobody has tested a newer one. Among the 20 highest-ranked models on the capability index, these have no score yet:

  1. 1 Gemini 3 Pro Zero-shot, reasoning 97.3%
  2. 2 Gemini 3 Flash Zero-shot, reasoning 96.3%
  3. 3 GPT-5.2 Zero-shot, reasoning 94.2%
  4. 4 GPT-5 Zero-shot, reasoning 93.5%
  5. 5 GPT-5 mini Zero-shot, reasoning 92%
  6. 6 Qwen3-VL-32B-Thinking Zero-shot, reasoning 84.2%
  7. 7 GLM-4.6V Zero-shot, reasoning 82.1%
  8. 8 Qwen3-VL-30B-A3B-Thinking Zero-shot, reasoning 81.5%
  9. 9 Qwen3-VL-8B-Thinking Zero-shot, reasoning 76.9%
  10. 10 Qwen3-VL-32B-Instruct Zero-shot, instruct 76%
  11. 11 Ministral 3 14B Zero-shot, reasoning 74.9%
  12. 12 Qwen3-VL-30B-A3B-Instruct Zero-shot, instruct 70%
  13. 13 Qwen3-VL-4B-Thinking Zero-shot, reasoning 68%
  14. 14 Ministral-3-8B-Reasoning-2512 Zero-shot, reasoning 67%
  15. 15 Hulu-Med-32B Zero-shot, instruct 66.7%
  16. 16 GLM-4.6V-Flash Zero-shot, reasoning 66.5%
  17. 17 Lingshu-32B Zero-shot, instruct 65.4%
  18. 18 InternVL3.5-38B Zero-shot, instruct 64.9%
  19. 19 InternVL3.5-30B-A3B Zero-shot, instruct 63.9%
  20. 20 Qwen2.5-VL-32B Zero-shot, instruct 62%
  21. 21 Hulu-Med-14B Zero-shot, instruct 60.9%
  22. 22 Ministral-3-14B-Instruct-2512 Zero-shot, instruct 60.1%
  23. 23 Gemma 3 12B Zero-shot, instruct 59.5%
  24. 24 MedGemma-27B-IT Zero-shot, instruct 58.8%
  25. 25 Gemma 3 27B Zero-shot, instruct 58.5%
  26. 26 InternVL3.5-14B Zero-shot, instruct 55.5%
  27. 27 Ministral 3 8B Instruct Zero-shot, instruct 55.3%
  28. 28 Qwen3-VL-8B-Instruct Zero-shot, instruct 54.3%
  29. 29 VARCO-VISION-2.0-14B Zero-shot, instruct 52.1%
  30. 30 Ministral 3 3B Reasoning Zero-shot, reasoning 50.7%
  31. 31 Hulu-Med-7B Zero-shot, instruct 49.6%
  32. 32 InternVL3.5-8B Zero-shot, instruct 49.2%
  33. 33 Hulu-Med-4B Zero-shot, instruct 48.8%
  34. 34 A.X-4.0-VL-Light Zero-shot, instruct 48.7%
  35. 35 Lingshu-7B Zero-shot, instruct 48.4%
  36. 36 MedGemma-1.5-4B-IT Zero-shot, instruct 48.4%
  37. 37 Qwen2.5-VL-7B Zero-shot, instruct 46.5%
  38. 38 Qwen3-VL-4B-Instruct Zero-shot, instruct 46.3%
  39. 39 Qwen3-VL-2B-Thinking Zero-shot, reasoning 43.9%
  40. 40 Ministral 3 3B Instruct Zero-shot, instruct 43.6%
  41. 41 InternVL3.5-4B Zero-shot, instruct 41.8%
  42. 42 Kimi-VL-A3B-Thinking Zero-shot, reasoning 39.1%
  43. 43 MedGemma-4B-IT Zero-shot, instruct 38.8%
  44. 44 Kanana-1.5-V-3B-Instruct Zero-shot, instruct 35.8%
  45. 45 Gemma 3 4B Zero-shot, instruct 32.9%
  46. 46 Qwen3-VL-2B-Instruct Zero-shot, instruct 32.1%
  47. 47 VARCO-VISION-2.0-1.7B Zero-shot, instruct 31.3%
  48. 48 InternVL3.5-2B Zero-shot, instruct 30.6%
  49. 49 HyperCLOVAX-SEED-Vision-Instruct-3B Zero-shot, instruct 30.3%
  50. 50 Kimi-VL-A3B-Instruct Zero-shot, instruct 28.9%
  51. 51 InternVL3.5-1B Zero-shot, instruct 22.1%

Source: Official KorMedMCQA-Mixed leaderboard

  1. 1 Gemini 3 Pro Zero-shot, reasoning 97.3%
  2. 2 Gemini 3 Flash Zero-shot, reasoning 96.3%
  3. 3 GPT-5.2 Zero-shot, reasoning 94.2%
  4. 4 GPT-5 Zero-shot, reasoning 93.5%
  5. 5 GPT-5 mini Zero-shot, reasoning 92%
  6. 6 Qwen3-VL-32B-Thinking Zero-shot, reasoning 84.2%
  7. 7 GLM-4.6V Zero-shot, reasoning 82.1%
  8. 8 Qwen3-VL-30B-A3B-Thinking Zero-shot, reasoning 81.5%
  9. 9 Qwen3-VL-8B-Thinking Zero-shot, reasoning 76.9%
  10. 10 Qwen3-VL-32B-Instruct Zero-shot, instruct 76%
  11. 11 Ministral 3 14B Zero-shot, reasoning 74.9%
  12. 12 Qwen3-VL-30B-A3B-Instruct Zero-shot, instruct 70%
  13. 13 Qwen3-VL-4B-Thinking Zero-shot, reasoning 68%
  14. 14 GLM-4.6V-Flash Zero-shot, reasoning 66.5%
  15. 15 InternVL3.5-38B Zero-shot, instruct 64.9%
  16. 16 InternVL3.5-30B-A3B Zero-shot, instruct 63.9%
  17. 17 Qwen2.5-VL-32B Zero-shot, instruct 62%
  18. 18 Gemma 3 12B Zero-shot, instruct 59.5%
  19. 19 Gemma 3 27B Zero-shot, instruct 58.5%
  20. 20 InternVL3.5-14B Zero-shot, instruct 55.5%
  21. 21 Ministral 3 8B Instruct Zero-shot, instruct 55.3%
  22. 22 Qwen3-VL-8B-Instruct Zero-shot, instruct 54.3%
  23. 23 Ministral 3 3B Reasoning Zero-shot, reasoning 50.7%
  24. 24 InternVL3.5-8B Zero-shot, instruct 49.2%
  25. 25 Qwen2.5-VL-7B Zero-shot, instruct 46.5%
  26. 26 Qwen3-VL-4B-Instruct Zero-shot, instruct 46.3%
  27. 27 Qwen3-VL-2B-Thinking Zero-shot, reasoning 43.9%
  28. 28 Ministral 3 3B Instruct Zero-shot, instruct 43.6%
  29. 29 InternVL3.5-4B Zero-shot, instruct 41.8%
  30. 30 Kimi-VL-A3B-Thinking Zero-shot, reasoning 39.1%
  31. 31 Gemma 3 4B Zero-shot, instruct 32.9%
  32. 32 Qwen3-VL-2B-Instruct Zero-shot, instruct 32.1%
  33. 33 Kimi-VL-A3B-Instruct Zero-shot, instruct 28.9%

Source: Official KorMedMCQA-Mixed leaderboard

#EntryScoreSettingDateSource
1 Gemini 3 Pro 97.3% Zero-shot, reasoning Nov 2025 Official leaderboard
2 Gemini 3 Flash 96.3% Zero-shot, reasoning Dec 2025 Official leaderboard
3 GPT-5.2 94.2% Zero-shot, reasoning Dec 2025 Official leaderboard
4 GPT-5 93.5% Zero-shot, reasoning Aug 2025 Official leaderboard
5 GPT-5 mini 92% Zero-shot, reasoning Aug 2025 Official leaderboard
6 Qwen3-VL-32B-Thinking 84.2% Zero-shot, reasoning Oct 2025 Official leaderboard
7 GLM-4.6V 82.1% Zero-shot, reasoning Dec 2025 Official leaderboard
8 Qwen3-VL-30B-A3B-Thinking 81.5% Zero-shot, reasoning Oct 2025 Official leaderboard
9 Qwen3-VL-8B-Thinking 76.9% Zero-shot, reasoning Oct 2025 Official leaderboard
10 Qwen3-VL-32B-Instruct 76% Zero-shot, instruct Oct 2025 Official leaderboard
11 Ministral 3 14B 74.9% Zero-shot, reasoning Dec 2025 Official leaderboard
12 Qwen3-VL-30B-A3B-Instruct 70% Zero-shot, instruct Oct 2025 Official leaderboard
13 Qwen3-VL-4B-Thinking 68% Zero-shot, reasoning Oct 2025 Official leaderboard
14 Ministral-3-8B-Reasoning-2512 8B 67% Zero-shot, reasoning Official leaderboard
15 Hulu-Med-32B 32B 66.7% Zero-shot, instruct Official leaderboard
16 GLM-4.6V-Flash 66.5% Zero-shot, reasoning Dec 2025 Official leaderboard
17 Lingshu-32B 32B 65.4% Zero-shot, instruct Official leaderboard
18 InternVL3.5-38B 64.9% Zero-shot, instruct Aug 2025 Official leaderboard
19 InternVL3.5-30B-A3B 63.9% Zero-shot, instruct Aug 2025 Official leaderboard
20 Qwen2.5-VL-32B 62% Zero-shot, instruct Mar 2025 Official leaderboard
21 Hulu-Med-14B 14B 60.9% Zero-shot, instruct Official leaderboard
22 Ministral-3-14B-Instruct-2512 14B 60.1% Zero-shot, instruct Official leaderboard
23 Gemma 3 12B 59.5% Zero-shot, instruct Mar 2025 Official leaderboard
24 MedGemma-27B-IT 27B 58.8% Zero-shot, instruct Official leaderboard
25 Gemma 3 27B 58.5% Zero-shot, instruct Mar 2025 Official leaderboard
26 InternVL3.5-14B 55.5% Zero-shot, instruct Aug 2025 Official leaderboard
27 Ministral 3 8B Instruct 55.3% Zero-shot, instruct Dec 2025 Official leaderboard
28 Qwen3-VL-8B-Instruct 54.3% Zero-shot, instruct Oct 2025 Official leaderboard
29 VARCO-VISION-2.0-14B 14B 52.1% Zero-shot, instruct Official leaderboard
30 Ministral 3 3B Reasoning 50.7% Zero-shot, reasoning Dec 2025 Official leaderboard
31 Hulu-Med-7B 7B 49.6% Zero-shot, instruct Official leaderboard
32 InternVL3.5-8B 49.2% Zero-shot, instruct Aug 2025 Official leaderboard
33 Hulu-Med-4B 4B 48.8% Zero-shot, instruct Official leaderboard
34 A.X-4.0-VL-Light 7B 48.7% Zero-shot, instruct Official leaderboard
35 Lingshu-7B 7B 48.4% Zero-shot, instruct Official leaderboard
36 MedGemma-1.5-4B-IT 4B 48.4% Zero-shot, instruct Official leaderboard
37 Qwen2.5-VL-7B 46.5% Zero-shot, instruct Jan 2025 Official leaderboard
38 Qwen3-VL-4B-Instruct 46.3% Zero-shot, instruct Oct 2025 Official leaderboard
39 Qwen3-VL-2B-Thinking 43.9% Zero-shot, reasoning Oct 2025 Official leaderboard
40 Ministral 3 3B Instruct 43.6% Zero-shot, instruct Dec 2025 Official leaderboard
41 InternVL3.5-4B 41.8% Zero-shot, instruct Aug 2025 Official leaderboard
42 Kimi-VL-A3B-Thinking 39.1% Zero-shot, reasoning Apr 2025 Official leaderboard
43 MedGemma-4B-IT 4B 38.8% Zero-shot, instruct Official leaderboard
44 Kanana-1.5-V-3B-Instruct 3B 35.8% Zero-shot, instruct Official leaderboard
45 Gemma 3 4B 32.9% Zero-shot, instruct Mar 2025 Official leaderboard
46 Qwen3-VL-2B-Instruct 32.1% Zero-shot, instruct Oct 2025 Official leaderboard
47 VARCO-VISION-2.0-1.7B 1.7B 31.3% Zero-shot, instruct Official leaderboard
48 InternVL3.5-2B 2B 30.6% Zero-shot, instruct Official leaderboard
49 HyperCLOVAX-SEED-Vision-Instruct-3B 3B 30.3% Zero-shot, instruct Official leaderboard
50 Kimi-VL-A3B-Instruct 28.9% Zero-shot, instruct Apr 2025 Official leaderboard
51 InternVL3.5-1B 1B 22.1% Zero-shot, instruct Official leaderboard
#ModelScoreSettingDateSource
1 Gemini 3 Pro 97.3% Zero-shot, reasoning Nov 2025 Official leaderboard
2 Gemini 3 Flash 96.3% Zero-shot, reasoning Dec 2025 Official leaderboard
3 GPT-5.2 94.2% Zero-shot, reasoning Dec 2025 Official leaderboard
4 GPT-5 93.5% Zero-shot, reasoning Aug 2025 Official leaderboard
5 GPT-5 mini 92% Zero-shot, reasoning Aug 2025 Official leaderboard
6 Qwen3-VL-32B-Thinking 84.2% Zero-shot, reasoning Oct 2025 Official leaderboard
7 GLM-4.6V 82.1% Zero-shot, reasoning Dec 2025 Official leaderboard
8 Qwen3-VL-30B-A3B-Thinking 81.5% Zero-shot, reasoning Oct 2025 Official leaderboard
9 Qwen3-VL-8B-Thinking 76.9% Zero-shot, reasoning Oct 2025 Official leaderboard
10 Qwen3-VL-32B-Instruct 76% Zero-shot, instruct Oct 2025 Official leaderboard
11 Ministral 3 14B 74.9% Zero-shot, reasoning Dec 2025 Official leaderboard
12 Qwen3-VL-30B-A3B-Instruct 70% Zero-shot, instruct Oct 2025 Official leaderboard
13 Qwen3-VL-4B-Thinking 68% Zero-shot, reasoning Oct 2025 Official leaderboard
14 GLM-4.6V-Flash 66.5% Zero-shot, reasoning Dec 2025 Official leaderboard
15 InternVL3.5-38B 64.9% Zero-shot, instruct Aug 2025 Official leaderboard
16 InternVL3.5-30B-A3B 63.9% Zero-shot, instruct Aug 2025 Official leaderboard
17 Qwen2.5-VL-32B 62% Zero-shot, instruct Mar 2025 Official leaderboard
18 Gemma 3 12B 59.5% Zero-shot, instruct Mar 2025 Official leaderboard
19 Gemma 3 27B 58.5% Zero-shot, instruct Mar 2025 Official leaderboard
20 InternVL3.5-14B 55.5% Zero-shot, instruct Aug 2025 Official leaderboard
21 Ministral 3 8B Instruct 55.3% Zero-shot, instruct Dec 2025 Official leaderboard
22 Qwen3-VL-8B-Instruct 54.3% Zero-shot, instruct Oct 2025 Official leaderboard
23 Ministral 3 3B Reasoning 50.7% Zero-shot, reasoning Dec 2025 Official leaderboard
24 InternVL3.5-8B 49.2% Zero-shot, instruct Aug 2025 Official leaderboard
25 Qwen2.5-VL-7B 46.5% Zero-shot, instruct Jan 2025 Official leaderboard
26 Qwen3-VL-4B-Instruct 46.3% Zero-shot, instruct Oct 2025 Official leaderboard
27 Qwen3-VL-2B-Thinking 43.9% Zero-shot, reasoning Oct 2025 Official leaderboard
28 Ministral 3 3B Instruct 43.6% Zero-shot, instruct Dec 2025 Official leaderboard
29 InternVL3.5-4B 41.8% Zero-shot, instruct Aug 2025 Official leaderboard
30 Kimi-VL-A3B-Thinking 39.1% Zero-shot, reasoning Apr 2025 Official leaderboard
31 Gemma 3 4B 32.9% Zero-shot, instruct Mar 2025 Official leaderboard
32 Qwen3-VL-2B-Instruct 32.1% Zero-shot, instruct Oct 2025 Official leaderboard
33 Kimi-VL-A3B-Instruct 28.9% Zero-shot, instruct Apr 2025 Official leaderboard

Rankings by part

We read 6 parts from the benchmark’s own leaderboard in 3 groups: exam year, 2022 and 2023. Each ranks models on one part of the test. Gemini 3 Pro leads 5 of them, and Gemini 3 Flash leads the other one. Arrows show how far a model moved from its overall rank among the models we track.

KorMedMCQA-Mixed-2022

33 models we track have a score here.

  1. 1 Gemini 3 Pro 97.9%
  2. 2 Gemini 3 Flash 96.1%
  3. 3 GPT-5.2 93.6%
  4. 4 GPT-5 92.5%
  5. 5 GPT-5 mini 91.1%
  6. 6 Qwen3-VL-32B-Thinking 81.9%
  7. 7 Qwen3-VL-30B-A3B-Thinking ↑1 80.1%
  8. 8 GLM-4.6V ↓1 79.2%
  9. 9 Qwen3-VL-8B-Thinking 75.1%
  10. 10 Ministral 3 14B ↑1 75.1%

Number one over time

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

Google DeepMind
Own leaderboard Passed 85% in Feb 2026, by Gemini 3 Pro
Gemini 3 Pro 97.3%

Source: Official KorMedMCQA-Mixed leaderboard

GPT-5 on the official leaderboard passed 85% in August 2025

Models released before it came out already scored that high. All the scores we have, one dot each, so a model tested at several settings appears more than once.

Plot with other benchmarks
On the benchmark’s own leaderboard Best so far
0% 20% 40% 60% 80% 100% 2025 2026 Near the ceiling (85%) Gemini 3 Pro GPT-5 Qwen2.5-VL-32B Qwen2.5-VL-7B 0% 20% 40% 60% 80% 100% 2025 2026 Near the ceiling (85%) Gemini 3 Pro Qwen2.5-VL-32B

Source: Official KorMedMCQA-Mixed leaderboard

No lab model card reports it yet

We read the headline results of 540 launch posts from 57 labs, and none of them list this benchmark.

More in Law, finance and medicine

All 637 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%
  • 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. o1-preview 92.7%
  • Harvey LAB-AA v1.1 In use Artificial Analysis's run of Harvey's Legal Agent Benchmark on Harvey's newer private set of 120 tasks across 24 practice areas. A deliverable with a material hallucination scores zero, and the headline is the share of judges who find every rubric criterion passed. GPT-6.2 Astra 15%

Data audit

All 8 checks pass. Every benchmark

Catalog record 2 pass
  • Pass Task access All tasks are public.
  • Pass Confidence Checked against a primary source.
Scores 2 pass 3 notes
  • Pass Independent runs Scores from the official leaderboard (51 entries).
  • Pass Models scored 33 models we track, plus 18 other entries on its leaderboard.
  • Note Newest score The newest came 11 months ago, in December 2025, for Gemini 3 Flash. That’s expected for a saturated test.
  • Note Frontier models None of the top 10 has a score, which is fine for a retired test. The strongest tested is GPT-5.2 (#51).
  • Note Headroom Saturated. The best is 97.3%, past 85% since August 2025.
Leaderboard import 3 pass
  • Pass Importer We read the board on kormedmcqa-v.github.io: 33 rows for models we track and 18 other entries at the last read.
  • Pass Last read November 20th, 2026.
  • Pass Read worked The last read worked.
Consistency 1 pass 1 note
  • Note Lab claims No model has both a lab number and an independent run.
  • Pass Known issues No published audit found problems.

How we check