Benchmark Almanac

HCSU-Bench

Doubao-1.5-Vision-Pro 37.3% tops its leaderboard GPT-5.2 34.6% leads the models we track
Not in model cards

Models match a Chinese calligraphy sample to one of eight styles. A second task asks models to describe the brushwork in fewer than 50 words.

The authors rank nine models by mean style accuracy across ink manuscripts and stone rubbings. They score written critiques separately.

Released
March 2026
Kind
General visual questions
Level
University
Scored
Checked automatically
Answers
Multiple choice
Tasks
All public
Measures
Capability
Subject
Arts
Language
Chinese
In and out
Text, Images → Text
Interaction
Single turn
Refresh
One fixed set
Run by
Anyone
In views
Vision
Strongest model tested
GPT-5.2 #51 on the capability index
Last model added
December 2025
Scores refreshed
November 20th, 2026
9
Entries on its own leaderboard, 7 models we track

Doubao-1.5-Vision-Pro tops the leaderboard with 37.3% GPT-5.2 leads with 34.6% 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. The best model we track is GPT-5.2, at 34.6%. 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 Doubao-1.5-Vision-Pro 37.3%
  2. 2 GPT-5.2 34.6%
  3. 3 GPT-4o Mar 2025 32.9%
  4. 4 Claude Sonnet 4.5 32.8%
  5. 5 Qwen3-VL-235B-A22B-Instruct 30.3%
  6. 6 Gemini 2.5 Pro Jun 2025 29.3%
  7. 7 Qwen2.5-VL-72B 24.9%
  8. 8 Qwen3-VL-30B-Instruct 21.8%
  9. 9 Qwen2.5-VL-7B 20.3%

Source: Official HCSU-Bench leaderboard

#EntryScoreSettingDateSource
1 Doubao-1.5-Vision-Pro closed weights 37.3% Official leaderboard
2 GPT-5.2 34.6% Dec 2025 Official leaderboard
3 GPT-4o Mar 2025 32.9% Mar 2025 Official leaderboard
4 Claude Sonnet 4.5 32.8% Sep 2025 Official leaderboard
5 Qwen3-VL-235B-A22B-Instruct 30.3% Sep 2025 Official leaderboard
6 Gemini 2.5 Pro Jun 2025 29.3% Jun 2025 Official leaderboard
7 Qwen2.5-VL-72B 24.9% Jan 2025 Official leaderboard
8 Qwen3-VL-30B-Instruct open weights 21.8% Official leaderboard
9 Qwen2.5-VL-7B 20.3% Jan 2025 Official leaderboard
#ModelScoreSettingDateSource
1 GPT-5.2 34.6% Dec 2025 Official leaderboard
2 GPT-4o Mar 2025 32.9% Mar 2025 Official leaderboard
3 Claude Sonnet 4.5 32.8% Sep 2025 Official leaderboard
4 Qwen3-VL-235B-A22B-Instruct 30.3% Sep 2025 Official leaderboard
5 Gemini 2.5 Pro Jun 2025 29.3% Jun 2025 Official leaderboard
6 Qwen2.5-VL-72B 24.9% Jan 2025 Official leaderboard
7 Qwen2.5-VL-7B 20.3% Jan 2025 Official leaderboard

Rankings by part

We read 2 parts from the benchmark’s own leaderboard: Bei (Stone Rubbing) Acc and Tie (Ink Manuscript) Acc. Each ranks models on one part of the test. GPT-5.2 leads both. Arrows show how far a model moved from its overall rank among the models we track.

Bei (Stone Rubbing) Acc

7 models we track have a score here.

  1. 1 GPT-5.2 35.3%
  2. 2 Claude Sonnet 4.5 ↑1 34.3%
  3. 3 Qwen3-VL-235B-A22B-Instruct ↑1 34.1%
  4. 4 GPT-4o Mar 2025 ↓2 32.3%
  5. 5 Gemini 2.5 Pro Jun 2025 31.5%
  6. 6 Qwen2.5-VL-72B 26.6%
  7. 7 Qwen2.5-VL-7B 21.3%

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 Mar 2026
GPT-5.2 34.6%

Source: Official HCSU-Bench 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 Best so far
0% 20% 40% 60% 80% 100% 2025 2026 Near the ceiling (85%) GPT-5.2 GPT-4o Qwen2.5-VL-72B 0% 20% 40% 60% 80% 100% 2025 2026 Near the ceiling (85%) GPT-5.2 GPT-4o

Source: Official HCSU-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

9 of 10 checks pass, 1 warning. Look first at the Newest score check. The newest came 11 months ago, in December 2025, for GPT-5.2. 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 official leaderboard (9 entries).
  • Pass Models scored 7 models we track, plus 2 other entries on its leaderboard.
  • Warning Newest score The newest came 11 months ago, in December 2025, for GPT-5.2.
  • Note Frontier models None of the top 10 has a score. The strongest tested is GPT-5.2 (#51).
  • Pass Headroom The best is 34.6%, under 85%, so it still separates models.
Leaderboard import 3 pass
  • Pass Importer We read the board on github.com: 7 rows for models we track and 2 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