Benchmark Almanac

NOTA

Gemini-pro-vision 18.4% tops its leaderboard MiniCPM-Llama3-V 2.5 17.8% leads the models we track
Not in model cards

Models extract facts from sheet music, convert score images to ABC notation, and analyze musical structure. The authors report extraction rates and separate transcription distances.

The authors report 20,961 test examples across tasks. Microsoft Research later ran the image transcription task and reported an edit distance of 18.39.

Made by
Mingni Tang and colleagues
Released
June 2024
Kind
Documents, charts and text in images
Level
University
Scored
Checked automatically
Answers
Short answer
Tasks
All public
Measures
Capability
Subject
Arts
Language
English
In and out
Images, Text → Text
Interaction
Single turn
Refresh
One fixed set
Run by
Anyone
In views
Vision
Last model added
May 2024
Scores refreshed
November 20th, 2026
17
Entries in its paper, 10 models we track

Gemini-pro-vision tops the leaderboard with 18.4% MiniCPM-Llama3-V 2.5 leads with 17.8% among models released by May 2024

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. The best model we track is MiniCPM-Llama3-V 2.5, at 17.8%. This view shows only models we track, each at its best.

The newest model with a score here came out in May 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 Gemini-pro-vision 18.4%
  2. 2 MiniCPM-Llama3-V 2.5 17.8%
  3. 3 NotaGPT-7B 17.2%
  4. 4 InternVL-Chat-V1-5 16.9%
  5. 5 GPT4o (snapshot unspecified) 16.4%
  6. 6 GPT-4V 16%
  7. 7 VisualGLM-6B 15.8%
  8. 8 Yi-VL-34B 15.6%
  9. 9 instructblip-vicuna-7b 15.6%
  10. 10 InstructBLIP FLAN-T5-XL 15%
  11. 11 Yi-VL-6B 14.8%
  12. 12 instructblip-vicuna-13b 14.3%
  13. 13 InstructBLIP FLAN-T5-XXL 13.1%
  14. 14 Qwen-VL-Chat 13.1%
  15. 15 DeepSeek-VL-7B-base 12.9%
  16. 16 Qwen-VL 11.9%
  17. 17 InternVL-14B-224px 4.35%

Source: The NOTA paper

#EntryScoreSettingDateSource
1 Gemini-pro-vision 18.4% Official leaderboard
2 MiniCPM-Llama3-V 2.5 17.8% May 2024 Official leaderboard
3 NotaGPT-7B 17.2% Official leaderboard
4 InternVL-Chat-V1-5 16.9% Apr 2024 Official leaderboard
5 GPT4o (snapshot unspecified) 16.4% Official leaderboard
6 GPT-4V 16% Nov 2023 Official leaderboard
7 VisualGLM-6B 15.8% Official leaderboard
8 Yi-VL-34B 15.6% Dec 2023 Official leaderboard
9 instructblip-vicuna-7b 15.6% May 2023 Official leaderboard
10 InstructBLIP FLAN-T5-XL 15% May 2023 Official leaderboard
11 Yi-VL-6B 14.8% Dec 2023 Official leaderboard
12 instructblip-vicuna-13b 14.3% Dec 2023 Official leaderboard
13 InstructBLIP FLAN-T5-XXL 13.1% Jun 2023 Official leaderboard
14 Qwen-VL-Chat 13.1% Aug 2023 Official leaderboard
15 DeepSeek-VL-7B-base 12.9% Official leaderboard
16 Qwen-VL 11.9% Official leaderboard
17 InternVL-14B-224px 4.35% Official leaderboard
#ModelScoreSettingDateSource
1 MiniCPM-Llama3-V 2.5 17.8% May 2024 Official leaderboard
2 InternVL-Chat-V1-5 16.9% Apr 2024 Official leaderboard
3 GPT-4V 16% Nov 2023 Official leaderboard
4 Yi-VL-34B 15.6% Dec 2023 Official leaderboard
5 instructblip-vicuna-7b 15.6% May 2023 Official leaderboard
6 InstructBLIP FLAN-T5-XL 15% May 2023 Official leaderboard
7 Yi-VL-6B 14.8% Dec 2023 Official leaderboard
8 instructblip-vicuna-13b 14.3% Dec 2023 Official leaderboard
9 InstructBLIP FLAN-T5-XXL 13.1% Jun 2023 Official leaderboard
10 Qwen-VL-Chat 13.1% Aug 2023 Official leaderboard

Number one over time

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

OpenBMB (ModelBest)
Benchmark paper Not tested since Jun 2024

Source: The NOTA 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 Best so far
0% 20% 40% 60% 80% 100% 2023 2024 Near the ceiling (85%) MiniCPM-Llama3-V 2.5 GPT-4V instructblip-vicuna-7b 0% 20% 40% 60% 80% 100% 2023 2024 Near the ceiling (85%) MiniCPM-Llama3-V 2.5 instructblip-vicuna-7b

Source: Official NOTA 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

9 of 10 checks pass, 1 warning. Look first at the Newest score check. The newest came 2 years ago, in May 2024, for MiniCPM-Llama3-V 2.5. Every benchmark

Catalog record 2 pass
  • Pass Task access All tasks are public.
  • Pass Confidence Checked against a primary source.
Scores 1 warning 3 pass 1 note
  • Pass Independent runs Scores from the benchmark’s paper (17 entries).
  • Pass Models scored 10 models we track, plus 7 other entries on its leaderboard.
  • Warning Newest score The newest came 2 years ago, in May 2024, for MiniCPM-Llama3-V 2.5.
  • Note Frontier models None of its models are on the capabilities index.
  • Pass Headroom The best is 17.8%, under 85%, so it still separates models.
Leaderboard import 3 pass
  • Pass Importer We read the board on aclanthology.org: 10 rows for models we track and 7 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