Benchmark Almanac

CMMU

GPT-4V 30.9 leads since January 2024
Not in model cards

Models answer Chinese multimodal questions across seven school subjects.

The publisher reports model results.

Made by
BAAI FlagEval
Released
January 2024
Kind
General visual questions
Level
School
Scored
Checked automatically
Answers
Short answer
Tasks
All public
Measures
Capability
Language
Chinese
In and out
Text, Images → Text
Interaction
Single turn
Refresh
One fixed set
Run by
Anyone
Last model added
January 2024
Scores refreshed
November 20th, 2026
10
Entries on its own leaderboard, 5 models we track

GPT-4V leads with 30.9 among models released by January 2024 GPT-4V leads with 30.9 among models released by January 2024

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. Every run is in the next chart. 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 January 2024, and nobody has tested a newer one. Among the 20 highest-ranked models on the capability index, these have no score yet:

  1. 1 GPT-4V Fill-in-the-blank grading by GPT-4 30.9
  2. 2 Qwen-VL-Plus Fill-in-the-blank grading by GPT-4 27.7
  3. 3 Gemini 1.0 Pro Fill-in-the-blank grading by GPT-4 22.5
  4. 4 Intern-XComposer-7b Fill-in-the-blank grading by GPT-4 18.4
  5. 5 Qwen-VL-Chat-7b Fill-in-the-blank grading by GPT-4 12.1
  6. 6 LLaVA-v1.5-13B Fill-in-the-blank grading by GPT-4 12.0
  7. 7 mPLUG-Owl2-7b Fill-in-the-blank grading by GPT-4 8.58
  8. 8 ShareGPT4V-7B Fill-in-the-blank grading by GPT-4 7.63
  9. 9 CogVLM-7b Fill-in-the-blank grading by GPT-4 4.90
  10. 10 InstructBLIP-13b Fill-in-the-blank grading by GPT-4 0.48

Source: Official CMMU leaderboard

#EntryScoreSettingDateSource
1 GPT-4V 30.9 Fill-in-the-blank grading by GPT-4 Nov 2023 Official leaderboard
2 Qwen-VL-Plus 27.7 Fill-in-the-blank grading by GPT-4 Jan 2024 Official leaderboard
3 Gemini 1.0 Pro 22.5 Fill-in-the-blank grading by GPT-4 Dec 2023 Official leaderboard
4 Intern-XComposer-7b 18.4 Fill-in-the-blank grading by GPT-4 Official leaderboard
5 Qwen-VL-Chat-7b 12.1 Fill-in-the-blank grading by GPT-4 Official leaderboard
6 LLaVA-v1.5-13B 12.0 Fill-in-the-blank grading by GPT-4 Oct 2023 Official leaderboard
7 mPLUG-Owl2-7b 8.58 Fill-in-the-blank grading by GPT-4 Official leaderboard
8 ShareGPT4V-7B 7.63 Fill-in-the-blank grading by GPT-4 Nov 2023 Official leaderboard
9 CogVLM-7b 4.90 Fill-in-the-blank grading by GPT-4 Official leaderboard
10 InstructBLIP-13b 0.48 Fill-in-the-blank grading by GPT-4 Official leaderboard
#ModelScoreSettingDateSource
1 GPT-4V 30.9 Fill-in-the-blank grading by GPT-4 Nov 2023 Official leaderboard
2 Qwen-VL-Plus 27.7 Fill-in-the-blank grading by GPT-4 Jan 2024 Official leaderboard
3 Gemini 1.0 Pro 22.5 Fill-in-the-blank grading by GPT-4 Dec 2023 Official leaderboard
4 LLaVA-v1.5-13B 12.0 Fill-in-the-blank grading by GPT-4 Oct 2023 Official leaderboard
5 ShareGPT4V-7B 7.63 Fill-in-the-blank grading by GPT-4 Nov 2023 Official leaderboard

Scores at each effort level

Models the leaderboard ran at two or more effort levels. We don’t have enough cost data here to say what the extra effort costs.

Lowest effort to highest, light to dark
GPT-4V
Fill-in-the… Fill-in-the…
30.9
Qwen-VL-Plus
Fill-in-the… Fill-in-the…
27.7
Gemini 1.0 Pro
Fill-in-the… Fill-in-the…
22.5
LLaVA-v1.5-13B
Fill-in-the… Fill-in-the…
12.0
ShareGPT4V-7B
Fill-in-the…
7.63

Source: Official CMMU leaderboard

ModelEffortHarnessScoreCost per taskOutput tokens
GPT-4V Fill-in-the-blank grading by GPT-4 30.9
Qwen-VL-Plus Fill-in-the-blank grading by GPT-4 27.7
GPT-4V Fill-in-the-blank grading by rules 25.2
Gemini 1.0 Pro Fill-in-the-blank grading by GPT-4 22.5
Qwen-VL-Plus Fill-in-the-blank grading by rules 21.3
Gemini 1.0 Pro Fill-in-the-blank grading by rules 17.9
LLaVA-v1.5-13B Fill-in-the-blank grading by GPT-4 12.0
LLaVA-v1.5-13B Fill-in-the-blank grading by rules 8.06
ShareGPT4V-7B Fill-in-the-blank grading by GPT-4 7.63
ShareGPT4V-7B Fill-in-the-blank grading by rules 5.07

Scores at each effort level

Models the leaderboard ran at two or more effort levels. We don’t have enough cost data here to say what the extra effort costs.

Lowest effort to highest, light to dark
GPT-4V
Fill-in-the… Fill-in-the…
30.9
Qwen-VL-Plus
Fill-in-the… Fill-in-the…
27.7
Gemini 1.0 Pro
Fill-in-the… Fill-in-the…
22.5
LLaVA-v1.5-13B
Fill-in-the… Fill-in-the…
12.0
ShareGPT4V-7B
Fill-in-the…
7.63

Source: Official CMMU leaderboard

ModelEffortHarnessScoreCost per taskOutput tokens
GPT-4V Fill-in-the-blank grading by GPT-4 30.9
Qwen-VL-Plus Fill-in-the-blank grading by GPT-4 27.7
GPT-4V Fill-in-the-blank grading by rules 25.2
Gemini 1.0 Pro Fill-in-the-blank grading by GPT-4 22.5
Qwen-VL-Plus Fill-in-the-blank grading by rules 21.3
Gemini 1.0 Pro Fill-in-the-blank grading by rules 17.9
LLaVA-v1.5-13B Fill-in-the-blank grading by GPT-4 12.0
LLaVA-v1.5-13B Fill-in-the-blank grading by rules 8.06
ShareGPT4V-7B Fill-in-the-blank grading by GPT-4 7.63
ShareGPT4V-7B Fill-in-the-blank grading by rules 5.07

Number one over time

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

OpenAI
Own leaderboard Not tested since Jan 2024
GPT-4V 30.9

Source: Official CMMU leaderboard

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
On the benchmark’s own leaderboard
0.00 10.0 20.0 30.0 40.0 2024 0.00 10.0 20.0 30.0 40.0 2024

Source: Official CMMU 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 2 years ago, in January 2024, for Qwen-VL-Plus. Every benchmark

Catalog record 2 pass
  • Pass Task access All tasks are public.
  • Pass Confidence Checked against a primary source.
Scores 1 warning 2 pass 2 notes
  • Pass Independent runs Scores from the official leaderboard (10 entries).
  • Pass Models scored 5 models we track, plus 5 other entries on its leaderboard.
  • Warning Newest score The newest came 2 years ago, in January 2024, for Qwen-VL-Plus.
  • Note Frontier models None of its models are on the capabilities index.
  • Note Headroom Scored as a plain number, with no fixed ceiling.
Leaderboard import 3 pass
  • Pass Importer We read the board on github.com: 10 rows for models we track and 5 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