Benchmark Almanac

MCS-Bench

InternVL2_5-78B 49.3 leads since February 2025
Not in model cards

Models read seal inscriptions, identify calligraphy, and answer questions about Chinese paintings and artifacts. The overall score averages task scores.

The authors compare 37 models on 23 multiple-choice and 22 open-ended tasks. Institutions must apply for the full dataset.

Released
February 2025
Kind
General visual questions
Level
Expert
Scored
Checked automatically
Answers
Multiple choice
Tasks
Kept private
Measures
Capability
Language
Chinese
In and out
Text, Images → Text
Interaction
Single turn
Refresh
One fixed set
Run by
A host
In views
Vision
Strongest model tested
Gemini 2.0 Flash Feb 2025 #147 on the capability index
Last model added
March 2025
Scores refreshed
November 20th, 2026
37
Entries in its paper, 36 models we track

InternVL2_5-78B leads with 49.3 among models released by March 2025 InternVL2_5-78B leads with 49.3 among models released by March 2025

Best score per entry from the benchmark's paper, read on November 20th, 2026. A paper’s table doesn’t add newer models. 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 March 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 InternVL2_5-78B 49.3
  2. 2 InternVL2.5-38B 46.5
  3. 3 Gemini 2.0 Flash Feb 2025 46.2
  4. 4 InternVL2.5-26B 45.9
  5. 5 GPT-4o Mar 2025 44.7
  6. 6 Qwen2-VL-72B-Instruct 44.6
  7. 7 Claude 3.5 Sonnet 44.5
  8. 8 Gemini 1.5 Pro Sep 2024 43.9
  9. 9 InternVL2.5-8B 43.6
  10. 10 InternVL2.5-4B 43.3
  11. 11 InternVL2-26B 42.7
  12. 12 QVQ-72B-Preview 42.5
  13. 13 GLM-4V-9B 41.8
  14. 14 InternVL2-8B 41.0
  15. 15 Qwen2-VL-7B-Instruct 40.6
  16. 16 MiniCPM-V-2_6 39.1
  17. 17 InternVL2.5-1B 38.8
  18. 18 InternVL2.5-2B 38.8
  19. 19 Qwen2-VL-2B-Instruct 35.6
  20. 20 InternVL2-4B 34.8
  21. 21 DeepSeek-VL2-Tiny 34.2
  22. 22 Qwen-VL-Chat 33.5
  23. 23 Ovis1.6-Gemma2-9B 32.5
  24. 24 MiniCPM-V 2.0 32.5
  25. 25 MiniCPM-Llama3-V 2.5 32.4
  26. 26 DeepSeek-VL2 30.5
  27. 27 MiniCPM-V 30.4
  28. 28 Ovis1.5-Gemma2-9B 30.2
  29. 29 DeepSeek-VL2-Small 26.3
  30. 30 llava-v1.6-vicuna-7b 24.5
  31. 31 LLaVA-v1.5-13B 24.1
  32. 32 llava-v1.6-vicuna-13b 23.6
  33. 33 llava-v1.6-mistral-7b 23.4
  34. 34 llava-v1.5-7b 23.3
  35. 35 Molmo-7B-D 21.2
  36. 36 Chinese-LLaVA-CLLaMA2 21.0
  37. 37 Molmo-7B-O 20.2

Source: The MCS-Bench paper

Source: The MCS-Bench paper

#EntryScoreSettingDateSource
1 InternVL2_5-78B 49.3 Dec 2024 Official leaderboard
2 InternVL2.5-38B 46.5 Dec 2024 Official leaderboard
3 Gemini 2.0 Flash Feb 2025 46.2 Feb 2025 Official leaderboard
4 InternVL2.5-26B 45.9 Dec 2024 Official leaderboard
5 GPT-4o Mar 2025 44.7 Mar 2025 Official leaderboard
6 Qwen2-VL-72B-Instruct 44.6 Aug 2024 Official leaderboard
7 Claude 3.5 Sonnet 44.5 Jun 2024 Official leaderboard
8 Gemini 1.5 Pro Sep 2024 43.9 Sep 2024 Official leaderboard
9 InternVL2.5-8B 43.6 Dec 2024 Official leaderboard
10 InternVL2.5-4B 43.3 Dec 2024 Official leaderboard
11 InternVL2-26B 42.7 Jul 2024 Official leaderboard
12 QVQ-72B-Preview 42.5 Dec 2024 Official leaderboard
13 GLM-4V-9B 41.8 Jun 2024 Official leaderboard
14 InternVL2-8B 41.0 Jul 2024 Official leaderboard
15 Qwen2-VL-7B-Instruct 40.6 Aug 2024 Official leaderboard
16 MiniCPM-V-2_6 39.1 Aug 2024 Official leaderboard
17 InternVL2.5-1B 38.8 Dec 2024 Official leaderboard
18 InternVL2.5-2B 38.8 Dec 2024 Official leaderboard
19 Qwen2-VL-2B-Instruct 35.6 Aug 2024 Official leaderboard
20 InternVL2-4B 34.8 Jul 2024 Official leaderboard
21 DeepSeek-VL2-Tiny 34.2 Dec 2024 Official leaderboard
22 Qwen-VL-Chat 33.5 Aug 2023 Official leaderboard
23 Ovis1.6-Gemma2-9B 32.5 Sep 2024 Official leaderboard
24 MiniCPM-V 2.0 32.5 Apr 2024 Official leaderboard
25 MiniCPM-Llama3-V 2.5 32.4 May 2024 Official leaderboard
26 DeepSeek-VL2 30.5 Dec 2024 Official leaderboard
27 MiniCPM-V 30.4 Jan 2024 Official leaderboard
28 Ovis1.5-Gemma2-9B 30.2 Jul 2024 Official leaderboard
29 DeepSeek-VL2-Small 26.3 Dec 2024 Official leaderboard
30 llava-v1.6-vicuna-7b 24.5 Jan 2024 Official leaderboard
31 LLaVA-v1.5-13B 24.1 Oct 2023 Official leaderboard
32 llava-v1.6-vicuna-13b 23.6 Jan 2024 Official leaderboard
33 llava-v1.6-mistral-7b 23.4 Jan 2024 Official leaderboard
34 llava-v1.5-7b 23.3 Oct 2023 Official leaderboard
35 Molmo-7B-D 21.2 Sep 2024 Official leaderboard
36 Chinese-LLaVA-CLLaMA2 open weights 21.0 Official leaderboard
37 Molmo-7B-O 20.2 Sep 2024 Official leaderboard
#ModelScoreSettingDateSource
1 InternVL2_5-78B 49.3 Dec 2024 Official leaderboard
2 InternVL2.5-38B 46.5 Dec 2024 Official leaderboard
3 Gemini 2.0 Flash Feb 2025 46.2 Feb 2025 Official leaderboard
4 InternVL2.5-26B 45.9 Dec 2024 Official leaderboard
5 GPT-4o Mar 2025 44.7 Mar 2025 Official leaderboard
6 Qwen2-VL-72B-Instruct 44.6 Aug 2024 Official leaderboard
7 Claude 3.5 Sonnet 44.5 Jun 2024 Official leaderboard
8 Gemini 1.5 Pro Sep 2024 43.9 Sep 2024 Official leaderboard
9 InternVL2.5-8B 43.6 Dec 2024 Official leaderboard
10 InternVL2.5-4B 43.3 Dec 2024 Official leaderboard
11 InternVL2-26B 42.7 Jul 2024 Official leaderboard
12 QVQ-72B-Preview 42.5 Dec 2024 Official leaderboard
13 GLM-4V-9B 41.8 Jun 2024 Official leaderboard
14 InternVL2-8B 41.0 Jul 2024 Official leaderboard
15 Qwen2-VL-7B-Instruct 40.6 Aug 2024 Official leaderboard
16 MiniCPM-V-2_6 39.1 Aug 2024 Official leaderboard
17 InternVL2.5-1B 38.8 Dec 2024 Official leaderboard
18 InternVL2.5-2B 38.8 Dec 2024 Official leaderboard
19 Qwen2-VL-2B-Instruct 35.6 Aug 2024 Official leaderboard
20 InternVL2-4B 34.8 Jul 2024 Official leaderboard
21 DeepSeek-VL2-Tiny 34.2 Dec 2024 Official leaderboard
22 Qwen-VL-Chat 33.5 Aug 2023 Official leaderboard
23 Ovis1.6-Gemma2-9B 32.5 Sep 2024 Official leaderboard
24 MiniCPM-V 2.0 32.5 Apr 2024 Official leaderboard
25 MiniCPM-Llama3-V 2.5 32.4 May 2024 Official leaderboard
26 DeepSeek-VL2 30.5 Dec 2024 Official leaderboard
27 MiniCPM-V 30.4 Jan 2024 Official leaderboard
28 Ovis1.5-Gemma2-9B 30.2 Jul 2024 Official leaderboard
29 DeepSeek-VL2-Small 26.3 Dec 2024 Official leaderboard
30 llava-v1.6-vicuna-7b 24.5 Jan 2024 Official leaderboard
31 LLaVA-v1.5-13B 24.1 Oct 2023 Official leaderboard
32 llava-v1.6-vicuna-13b 23.6 Jan 2024 Official leaderboard
33 llava-v1.6-mistral-7b 23.4 Jan 2024 Official leaderboard
34 llava-v1.5-7b 23.3 Oct 2023 Official leaderboard
35 Molmo-7B-D 21.2 Sep 2024 Official leaderboard
36 Molmo-7B-O 20.2 Sep 2024 Official leaderboard

Rankings by part

We read 54 parts from the benchmark’s paper, each ranking models on one part of the test, such as Ancient Chinese Text, Calligraphy and Painting. InternVL2_5-78B leads 20 of them, and 12 other models lead the rest. Arrows show how far a model moved from its overall rank among the models we track.

Ancient Chinese Text

36 models we track have a score here.

  1. 1 Gemini 2.0 Flash Feb 2025 ↑2 51.4
  2. 2 Gemini 1.5 Pro Sep 2024 ↑6 50.6
  3. 3 InternVL2_5-78B ↓2 42.9
  4. 4 Qwen2-VL-72B-Instruct ↑2 42.8
  5. 5 InternVL2.5-38B ↓3 39.5
  6. 6 Claude 3.5 Sonnet ↑1 38.5
  7. 7 Qwen2-VL-7B-Instruct ↑8 37.8
  8. 8 InternVL2.5-8B ↑1 36.9
  9. 9 InternVL2.5-26B ↓5 36.8
  10. 10 GPT-4o Mar 2025 ↓5 35.5

Number one over time

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

Shanghai AI Lab
Benchmark paper Not tested since Mar 2025

Source: The MCS-Bench paper

Scores over time

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

Plot with other benchmarks
In the benchmark’s paper
20.0 30.0 40.0 50.0 2024 2025 20.0 30.0 40.0 50.0 2024 2025

Source: Official MCS-Bench 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 Vision

All 636 in Vision
  • MMMU In use 11.5K college-level multimodal questions across 30 subjects. Multiple-choice and open accuracy. Qwen 3.6 Plus 86%
  • 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%
  • MathVista In use Math reasoning in visual contexts (charts, geometry, puzzles) from 31 datasets. Answer accuracy. Seed2.1 Pro 90.7%
  • 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%

Data audit

8 of 9 checks pass, 1 warning. Look first at the Newest score check. The newest came 20 months ago, in March 2025, for GPT-4o. Every benchmark

Catalog record 2 pass
  • Pass Task access The tasks are private.
  • Pass Confidence Checked against a primary source.
Scores 1 warning 2 pass 2 notes
  • Pass Independent runs Scores from the benchmark’s paper (37 entries).
  • Pass Models scored 36 models we track, plus 1 other entry on its leaderboard.
  • Warning Newest score The newest came 20 months ago, in March 2025, for GPT-4o.
  • Note Frontier models None of the top 10 has a score. The strongest tested is Gemini 2.0 Flash (#147).
  • Note Headroom Scored as a plain number, with no fixed ceiling.
Leaderboard import 3 pass
  • Pass Importer We read the board on github.com: 36 rows for models we track and 1 other entry 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