Benchmark Almanac

ImageNetV2

nasnetalarge 78.8% tops its leaderboard
Not in model cards

Models classify a newly collected set of photographs using the original ImageNet labels. The test measures how accuracy changes on fresh images.

Released
February 2019
Kind
General visual questions
Level
Everyday
Scored
Checked automatically
Answers
Short answer
Tasks
All public
Measures
Capability
Language
English
In and out
Images → Text
Interaction
Single turn
Items from
Real work and use
Run by
Anyone
Based on
ImageNet-1K
In views
Vision
67
Entries in its paper, 0 models we track

nasnetalarge tops the leaderboard with 78.8%

Best score per entry from the benchmark's paper, read on November 20th, 2026. A paper’s table doesn’t add newer models. The paper reports no overall accuracy, so the headline is the plain mean of top-1 accuracy on its three test sets. 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 nasnetalarge 78.8%
  2. 2 nasnet_large_tf 78.8%
  3. 3 pnasnet_large_tf 78.8%
  4. 4 pnasnet5large 78.8%
  5. 5 senet154 77.3%
  6. 6 polynet 77.3%
  7. 7 inceptionresnetv2 76.8%
  8. 8 inception_resnet_v2_tf 76.7%
  9. 9 inceptionv4 76.4%
  10. 10 se_resnext101_32x4d 76.4%
  11. 11 inception_v4_tf 76.3%
  12. 12 dpn107 75.4%
  13. 13 dpn131 75.3%
  14. 14 se_resnext50_32x4d 75.3%
  15. 15 dpn98 75.2%
  16. 16 dpn92 74.9%
  17. 17 se_resnet152 74.9%
  18. 18 xception 74.9%
  19. 19 se_resnet101 74.5%
  20. 20 resnext101_64x4d 74.3%
  21. 21 resnet152 74.2%
  22. 22 resnet_v2_152_tf 73.9%
  23. 23 resnext101_32x4d 73.8%
  24. 24 inception_v3_tf 73.7%
  25. 25 fbresnet152 73.6%
  26. 26 inceptionv3 73.3%
  27. 27 se_resnet50 73.3%
  28. 28 inception_v3 73.2%
  29. 29 densenet161 73.1%
  30. 30 resnet101 73.1%
  31. 31 resnet_v2_101_tf 72.9%
  32. 32 dpn68b 72.6%
  33. 33 densenet201 72.5%
  34. 34 resnet_v1_152_tf 72.4%
  35. 35 resnet_v1_101_tf 72.2%
  36. 36 cafferesnet101 71.8%
  37. 37 dpn68 71.6%
  38. 38 densenet169 71.4%
  39. 39 resnet50 71.4%
  40. 40 resnet_v2_50_tf 71%
  41. 41 resnet_v1_50_tf 70.5%
  42. 42 bninception 70.1%
  43. 43 vgg19_bn 70%
  44. 44 densenet121 69.8%
  45. 45 inception_v2_tf 69.8%
  46. 46 nasnetamobile 69.6%
  47. 47 pnasnet_mobile_tf 69.4%
  48. 48 resnet34 69.2%
  49. 49 vgg16_bn 69%
  50. 50 nasnet_mobile_tf 68.9%
  51. 51 vgg19 67.5%
  52. 52 vgg16 66.9%
  53. 53 vgg13_bn 66.6%
  54. 54 vgg_19_tf 66.5%
  55. 55 vgg_16_tf 66.5%
  56. 56 resnet18 65.6%
  57. 57 vgg11_bn 65.6%
  58. 58 inception_v1_tf 65.4%
  59. 59 mobilenet_v1_tf 65.3%
  60. 60 vgg13 65%
  61. 61 vgg11 63.7%
  62. 62 squeezenet1_1 53.8%
  63. 63 squeezenet1_0 53%
  64. 64 alexnet 51.2%
  65. 65 fv_64k 29.1%
  66. 66 fv_16k 23.3%
  67. 67 fv_4k 18%

Source: The ImageNetV2 paper

#EntryScoreSettingDateSource
1 nasnetalarge 78.8% Official leaderboard
2 nasnet_large_tf 78.8% Official leaderboard
3 pnasnet_large_tf 78.8% Official leaderboard
4 pnasnet5large 78.8% Official leaderboard
5 senet154 77.3% Official leaderboard
6 polynet 77.3% Official leaderboard
7 inceptionresnetv2 76.8% Official leaderboard
8 inception_resnet_v2_tf 76.7% Official leaderboard
9 inceptionv4 76.4% Official leaderboard
10 se_resnext101_32x4d 76.4% Official leaderboard
11 inception_v4_tf 76.3% Official leaderboard
12 dpn107 75.4% Official leaderboard
13 dpn131 75.3% Official leaderboard
14 se_resnext50_32x4d 75.3% Official leaderboard
15 dpn98 75.2% Official leaderboard
16 dpn92 74.9% Official leaderboard
17 se_resnet152 74.9% Official leaderboard
18 xception 74.9% Official leaderboard
19 se_resnet101 74.5% Official leaderboard
20 resnext101_64x4d 74.3% Official leaderboard
21 resnet152 74.2% Official leaderboard
22 resnet_v2_152_tf 73.9% Official leaderboard
23 resnext101_32x4d 73.8% Official leaderboard
24 inception_v3_tf 73.7% Official leaderboard
25 fbresnet152 73.6% Official leaderboard
26 inceptionv3 73.3% Official leaderboard
27 se_resnet50 73.3% Official leaderboard
28 inception_v3 73.2% Official leaderboard
29 densenet161 73.1% Official leaderboard
30 resnet101 73.1% Official leaderboard
31 resnet_v2_101_tf 72.9% Official leaderboard
32 dpn68b 72.6% Official leaderboard
33 densenet201 72.5% Official leaderboard
34 resnet_v1_152_tf 72.4% Official leaderboard
35 resnet_v1_101_tf 72.2% Official leaderboard
36 cafferesnet101 71.8% Official leaderboard
37 dpn68 71.6% Official leaderboard
38 densenet169 71.4% Official leaderboard
39 resnet50 71.4% Official leaderboard
40 resnet_v2_50_tf 71% Official leaderboard
41 resnet_v1_50_tf 70.5% Official leaderboard
42 bninception 70.1% Official leaderboard
43 vgg19_bn 70% Official leaderboard
44 densenet121 69.8% Official leaderboard
45 inception_v2_tf 69.8% Official leaderboard
46 nasnetamobile 69.6% Official leaderboard
47 pnasnet_mobile_tf 69.4% Official leaderboard
48 resnet34 69.2% Official leaderboard
49 vgg16_bn 69% Official leaderboard
50 nasnet_mobile_tf 68.9% Official leaderboard
51 vgg19 67.5% Official leaderboard
52 vgg16 66.9% Official leaderboard
53 vgg13_bn 66.6% Official leaderboard
54 vgg_19_tf 66.5% Official leaderboard
55 vgg_16_tf 66.5% Official leaderboard
56 resnet18 65.6% Official leaderboard
57 vgg11_bn 65.6% Official leaderboard
58 inception_v1_tf 65.4% Official leaderboard
59 mobilenet_v1_tf 65.3% Official leaderboard
60 vgg13 65% Official leaderboard
61 vgg11 63.7% Official leaderboard
62 squeezenet1_1 53.8% Official leaderboard
63 squeezenet1_0 53% Official leaderboard
64 alexnet 51.2% Official leaderboard
65 fv_64k 29.1% Official leaderboard
66 fv_16k 23.3% Official leaderboard
67 fv_4k 18% 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 (67 entries).
  • Pass Models scored 67 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: 67 entries at the last read, none a model we track.
  • 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