Benchmark Almanac

ImageNet-UA

DINOv2 ViT-large Patch14 27.7% tops its leaderboard
Not in model cards

Models classify images under distortions that differ from those used during training. The updated suite measures accuracy across 18 new distortions.

The Center for AI Safety maintains the updated evaluation and distinguishes it from the original implementation.

Released
July 2023
Kind
General visual questions
Level
Everyday
Scored
Checked automatically
Answers
Multiple choice
Tasks
All public
Measures
Capability
In and out
Images → Text
Interaction
Single turn
Items from
Another benchmark
Run by
Anyone
Based on
ImageNet-1K
In views
Vision
10
Entries in its paper, 0 models we track

DINOv2 ViT-large Patch14 tops the leaderboard with 27.7%

Best score per entry from the benchmark's paper, read on November 21st, 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.

  1. 1 DINOv2 ViT-large Patch14 27.7%
  2. 2 ViT-base Patch16, L_{\infty} (4/255) 25.8%
  3. 3 ConvNeXt-base, L_{\infty} (4/255) 22.3%
  4. 4 ConvNeXt-V2-large IN-1K+22K 19.2%
  5. 5 ConvNeXt-V2-huge IN-1K 17.7%
  6. 6 Swin-base IN-1K 15.2%
  7. 7 ResNet-50, L_{2} (5) 13.9%
  8. 8 ResNet-50, L_{\infty} (8/255) 10%
  9. 9 ResNet-50 + CutMix 6.10%
  10. 10 ResNet-50 1.60%

Source: The ImageNet-UA paper

#EntryScoreSettingDateSource
1 DINOv2 ViT-large Patch14 27.7% Official leaderboard
2 ViT-base Patch16, L_{\infty} (4/255) 25.8% Official leaderboard
3 ConvNeXt-base, L_{\infty} (4/255) 22.3% Official leaderboard
4 ConvNeXt-V2-large IN-1K+22K 19.2% Official leaderboard
5 ConvNeXt-V2-huge IN-1K 17.7% Official leaderboard
6 Swin-base IN-1K 15.2% Official leaderboard
7 ResNet-50, L_{2} (5) 13.9% Official leaderboard
8 ResNet-50, L_{\infty} (8/255) 10% Official leaderboard
9 ResNet-50 + CutMix 6.10% Official leaderboard
10 ResNet-50 1.60% Official 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

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 benchmark’s paper (10 entries).
  • Pass Models scored 10 entries on its leaderboard, none a model we track.
  • Note Newest score No scores yet.
  • Note Frontier models No models scored yet.
  • Note Headroom No scores yet.
Leaderboard import 3 pass
  • Pass Importer We read the board on arxiv.org: 10 entries at the last read, none a model we track.
  • Pass Last read November 21st, 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