Benchmark Almanac

FGVC-Aircraft

google/siglip-so400m-patch14-384 78.5% tops its leaderboard EmbeddingGemma 2 39.7% leads the models we track
Not in model cards

Models identify aircraft variants from photographs.

The dataset also groups aircraft by family and manufacturer.

Released
June 2013
Kind
Detection and captioning
Level
University
Scored
Checked automatically
Answers
Prediction
Tasks
All public
Measures
Capability
Language
English
In and out
Images → Text
Interaction
Single turn
Items from
Real work and use
Refresh
One fixed set
Run by
Anyone
In views
Vision
Last model added
October 2026
Scores refreshed
November 20th, 2026
58
Entries on MTEB’s leaderboard, 1 model we track

google/siglip-so400m-patch14-384 tops the leaderboard with 78.5% EmbeddingGemma 2 leads with 39.7%

Best score per entry from MTEB’s 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 EmbeddingGemma 2, at 39.7%. This view shows only models we track, each at its best.

58 entries. Scroll sideways for the rest.

  1. 78.5 google/siglip-so400m-patch14-384
  2. 78.2 QuanSun/EVA02-CLIP-bigE-14-plus
  3. 75.4 google/siglip-large-patch16-384
  4. 74.4 QuanSun/EVA02-CLIP-bigE-14
  5. 71.2 google/siglip-large-patch16-256
  6. 70.4 facebook/dinov2-giant
  7. 70.1 google/siglip-base-patch16-512
  8. 70.0 laion/CLIP-ViT-bigG-14-laion2B-39B-b160k
  9. 69.3 LCO-Embedding/LCO-Embedding-Omni-7B
  10. 69.0 google/siglip-base-patch16-384
  11. 67.2 encord-team/ebind-full
  12. 66.1 google/siglip-base-patch16-256
  13. 66.1 google/siglip-base-patch16-224
  14. 66.0 laion/CLIP-ViT-g-14-laion2B-s34B-b88K
  15. 65.1 facebook/dinov2-large
  16. 65.1 laion/CLIP-ViT-H-14-laion2B-s32B-b79K
  17. 64.2 laion/CLIP-ViT-L-14-DataComp.XL-s13B-b90K
  18. 63.3 LCO-Embedding/LCO-Embedding-Omni-3B
  19. 62.7 facebook/dinov2-base
  20. 60.5 laion/CLIP-ViT-L-14-laion2B-s32B-b82K
  21. 59.9 google/siglip-base-patch16-256-multilingual
  22. 59.5 LanguageBind/LanguageBind_Omni
  23. 58.7 facebook/dinov2-small
  24. 56.2 openai/clip-vit-large-patch14
  25. 54.6 QuanSun/EVA02-CLIP-L-14
  26. 54.0 nomic-ai/nomic-embed-vision-v1.5
  27. 53.9 laion/CLIP-ViT-B-16-DataComp.XL-s13B-b90K
  28. 52.3 dnotitia/DNA-VL-STEER-2B
  29. 49.5 jinaai/jina-embeddings-v5-omni-small
  30. 49.3 laion/CLIP-ViT-B-32-DataComp.XL-s13B-b90K
  31. 47.1 laion/CLIP-ViT-B-32-laion2B-s34B-b79K
  32. 46.0 openai/clip-vit-base-patch16
  33. 40.2 kakaobrain/align-base
  34. 39.7 EmbeddingGemma 2
  35. 39.6 QuanSun/EVA02-CLIP-B-16
  36. 38.9 BAAI/bge-visualized-m3
  37. 38.1 openai/clip-vit-base-patch32
  38. 37.7 royokong/e5-v
  39. 35.9 BAAI/bge-visualized-base
  40. 35.2 Salesforce/blip2-opt-6.7b-coco
  41. 34.3 Salesforce/blip2-opt-2.7b
  42. 32.4 jinaai/jina-clip-v1
  43. 29.9 jinaai/jina-embeddings-v5-omni-nano
  44. 25.9 Haon-Chen/e5-omni-7B
  45. 24.0 Salesforce/blip-itm-base-coco
  46. 22.5 mongodb/voyage-multimodal-3
  47. 22.4 Salesforce/blip-itm-large-coco
  48. 22.1 TIGER-Lab/VLM2Vec-LoRA
  49. 22.0 TIGER-Lab/VLM2Vec-Full
  50. 21.8 Salesforce/blip-itm-large-flickr
  51. 20.6 Salesforce/blip-image-captioning-large
  52. 20.1 Salesforce/blip-itm-base-flickr
  53. 19.7 BidirLM/BidirLM-Omni-2.5B-Embedding
  54. 18.9 nyu-visionx/moco-v3-vit-b
  55. 18.6 nyu-visionx/moco-v3-vit-l
  56. 17.7 Tevatron/OmniEmbed-v0.1
  57. 17.3 Salesforce/blip-image-captioning-base
  58. 12.4 Haon-Chen/e5-omni-3B
  1. 1 google/siglip-so400m-patch14-384 Image classification 78.5%
  2. 2 QuanSun/EVA02-CLIP-bigE-14-plus Image classification 78.2%
  3. 3 google/siglip-large-patch16-384 Image classification 75.4%
  4. 4 QuanSun/EVA02-CLIP-bigE-14 Image classification 74.4%
  5. 5 google/siglip-large-patch16-256 Image classification 71.2%
  6. 6 facebook/dinov2-giant Image classification 70.4%
  7. 7 google/siglip-base-patch16-512 Image classification 70.1%
  8. 8 laion/CLIP-ViT-bigG-14-laion2B-39B-b160k Image classification 70%
  9. 9 LCO-Embedding/LCO-Embedding-Omni-7B Zero-shot classification 69.3%
  10. 10 google/siglip-base-patch16-384 Image classification 69%
  11. 11 encord-team/ebind-full Zero-shot classification 67.2%
  12. 12 google/siglip-base-patch16-256 Image classification 66.1%
  13. 13 google/siglip-base-patch16-224 Image classification 66.1%
  14. 14 laion/CLIP-ViT-g-14-laion2B-s34B-b88K Image classification 66%
  15. 15 facebook/dinov2-large Image classification 65.1%
  16. 16 laion/CLIP-ViT-H-14-laion2B-s32B-b79K Image classification 65.1%
  17. 17 laion/CLIP-ViT-L-14-DataComp.XL-s13B-b90K Image classification 64.2%
  18. 18 LCO-Embedding/LCO-Embedding-Omni-3B Zero-shot classification 63.3%
  19. 19 facebook/dinov2-base Image classification 62.7%
  20. 20 laion/CLIP-ViT-L-14-laion2B-s32B-b82K Image classification 60.5%
  21. 21 google/siglip-base-patch16-256-multilingual Image classification 59.9%
  22. 22 LanguageBind/LanguageBind_Omni Image classification 59.5%
  23. 23 facebook/dinov2-small Image classification 58.7%
  24. 24 openai/clip-vit-large-patch14 Image classification 56.2%
  25. 25 QuanSun/EVA02-CLIP-L-14 Image classification 54.6%
  26. 26 nomic-ai/nomic-embed-vision-v1.5 Image classification 54%
  27. 27 laion/CLIP-ViT-B-16-DataComp.XL-s13B-b90K Image classification 53.9%
  28. 28 dnotitia/DNA-VL-STEER-2B Image classification 52.3%
  29. 29 jinaai/jina-embeddings-v5-omni-small Image classification 49.5%
  30. 30 laion/CLIP-ViT-B-32-DataComp.XL-s13B-b90K Image classification 49.3%
  31. 31 laion/CLIP-ViT-B-32-laion2B-s34B-b79K Image classification 47.1%
  32. 32 openai/clip-vit-base-patch16 Image classification 46%
  33. 33 kakaobrain/align-base Image classification 40.2%
  34. 34 EmbeddingGemma 2 Zero-shot classification 39.7%
  35. 35 QuanSun/EVA02-CLIP-B-16 Image classification 39.6%
  36. 36 BAAI/bge-visualized-m3 Image classification 38.9%
  37. 37 openai/clip-vit-base-patch32 Image classification 38.1%
  38. 38 royokong/e5-v Image classification 37.7%
  39. 39 BAAI/bge-visualized-base Image classification 35.9%
  40. 40 Salesforce/blip2-opt-6.7b-coco Image classification 35.2%
  41. 41 Salesforce/blip2-opt-2.7b Image classification 34.3%
  42. 42 jinaai/jina-clip-v1 Image classification 32.4%
  43. 43 jinaai/jina-embeddings-v5-omni-nano Image classification 29.9%
  44. 44 Haon-Chen/e5-omni-7B Zero-shot classification 25.9%
  45. 45 Salesforce/blip-itm-base-coco Image classification 24%
  46. 46 mongodb/voyage-multimodal-3 Image classification 22.5%
  47. 47 Salesforce/blip-itm-large-coco Image classification 22.4%
  48. 48 TIGER-Lab/VLM2Vec-LoRA Image classification 22.1%
  49. 49 TIGER-Lab/VLM2Vec-Full Image classification 22%
  50. 50 Salesforce/blip-itm-large-flickr Image classification 21.8%
  51. 51 Salesforce/blip-image-captioning-large Image classification 20.6%
  52. 52 Salesforce/blip-itm-base-flickr Image classification 20.1%
  53. 53 BidirLM/BidirLM-Omni-2.5B-Embedding Zero-shot classification 19.7%
  54. 54 nyu-visionx/moco-v3-vit-b Image classification 18.9%
  55. 55 nyu-visionx/moco-v3-vit-l Image classification 18.6%
  56. 56 Tevatron/OmniEmbed-v0.1 Zero-shot classification 17.7%
  57. 57 Salesforce/blip-image-captioning-base Image classification 17.3%
  58. 58 Haon-Chen/e5-omni-3B Zero-shot classification 12.4%

Source: MTEB’s leaderboard

Source: MTEB’s leaderboard

#EntryScoreSettingDateLinkSource
1 google/siglip-so400m-patch14-384 open weights, 0.9B 78.5% Image classification huggingface.co Official leaderboard
2 QuanSun/EVA02-CLIP-bigE-14-plus open weights, 5B 78.2% Image classification huggingface.co Official leaderboard
3 google/siglip-large-patch16-384 open weights, 0.7B 75.4% Image classification huggingface.co Official leaderboard
4 QuanSun/EVA02-CLIP-bigE-14 open weights, 4.7B 74.4% Image classification huggingface.co Official leaderboard
5 google/siglip-large-patch16-256 open weights, 0.7B 71.2% Image classification huggingface.co Official leaderboard
6 facebook/dinov2-giant open weights, 1.1B 70.4% Image classification huggingface.co Official leaderboard
7 google/siglip-base-patch16-512 open weights, 0.2B 70.1% Image classification huggingface.co Official leaderboard
8 laion/CLIP-ViT-bigG-14-laion2B-39B-b160k open weights, 2.5B 70% Image classification huggingface.co Official leaderboard
9 LCO-Embedding/LCO-Embedding-Omni-7B open weights, 8.9B 69.3% Zero-shot classification huggingface.co Official leaderboard
10 google/siglip-base-patch16-384 open weights, 0.2B 69% Image classification huggingface.co Official leaderboard
11 encord-team/ebind-full open weights, 1.8B 67.2% Zero-shot classification huggingface.co Official leaderboard
12 google/siglip-base-patch16-256 open weights, 0.2B 66.1% Image classification huggingface.co Official leaderboard
13 google/siglip-base-patch16-224 open weights, 0.2B 66.1% Image classification huggingface.co Official leaderboard
14 laion/CLIP-ViT-g-14-laion2B-s34B-b88K open weights, 1.4B 66% Image classification huggingface.co Official leaderboard
15 facebook/dinov2-large open weights, 0.3B 65.1% Image classification huggingface.co Official leaderboard
16 laion/CLIP-ViT-H-14-laion2B-s32B-b79K open weights, 1B 65.1% Image classification huggingface.co Official leaderboard
17 laion/CLIP-ViT-L-14-DataComp.XL-s13B-b90K open weights, 0.4B 64.2% Image classification huggingface.co Official leaderboard
18 LCO-Embedding/LCO-Embedding-Omni-3B open weights, 4.7B 63.3% Zero-shot classification huggingface.co Official leaderboard
19 facebook/dinov2-base open weights, 0.1B 62.7% Image classification huggingface.co Official leaderboard
20 laion/CLIP-ViT-L-14-laion2B-s32B-b82K open weights, 0.4B 60.5% Image classification huggingface.co Official leaderboard
21 google/siglip-base-patch16-256-multilingual open weights, 0.4B 59.9% Image classification huggingface.co Official leaderboard
22 LanguageBind/LanguageBind_Omni open weights, 1.2B 59.5% Image classification github.com Official leaderboard
23 facebook/dinov2-small open weights, 0B 58.7% Image classification huggingface.co Official leaderboard
24 openai/clip-vit-large-patch14 open weights, 0.4B 56.2% Image classification huggingface.co Official leaderboard
25 QuanSun/EVA02-CLIP-L-14 open weights, 0.4B 54.6% Image classification huggingface.co Official leaderboard
26 nomic-ai/nomic-embed-vision-v1.5 open weights, 0.1B 54% Image classification huggingface.co Official leaderboard
27 laion/CLIP-ViT-B-16-DataComp.XL-s13B-b90K open weights, 0.2B 53.9% Image classification huggingface.co Official leaderboard
28 dnotitia/DNA-VL-STEER-2B open weights, 2.1B 52.3% Image classification huggingface.co Official leaderboard
29 jinaai/jina-embeddings-v5-omni-small open weights, 1.6B 49.5% Image classification huggingface.co Official leaderboard
30 laion/CLIP-ViT-B-32-DataComp.XL-s13B-b90K open weights, 0.2B 49.3% Image classification huggingface.co Official leaderboard
31 laion/CLIP-ViT-B-32-laion2B-s34B-b79K open weights, 0.2B 47.1% Image classification huggingface.co Official leaderboard
32 openai/clip-vit-base-patch16 open weights, 0.2B 46% Image classification huggingface.co Official leaderboard
33 kakaobrain/align-base open weights, 0.2B 40.2% Image classification huggingface.co Official leaderboard
34 EmbeddingGemma 2 39.7% Zero-shot classification Oct 2026 huggingface.co Official leaderboard
35 QuanSun/EVA02-CLIP-B-16 open weights, 0.1B 39.6% Image classification huggingface.co Official leaderboard
36 BAAI/bge-visualized-m3 open weights, 0.9B 38.9% Image classification huggingface.co Official leaderboard
37 openai/clip-vit-base-patch32 open weights, 0.2B 38.1% Image classification huggingface.co Official leaderboard
38 royokong/e5-v open weights, 8.4B 37.7% Image classification huggingface.co Official leaderboard
39 BAAI/bge-visualized-base open weights, 0.2B 35.9% Image classification huggingface.co Official leaderboard
40 Salesforce/blip2-opt-6.7b-coco open weights, 7.8B 35.2% Image classification huggingface.co Official leaderboard
41 Salesforce/blip2-opt-2.7b open weights, 3.7B 34.3% Image classification huggingface.co Official leaderboard
42 jinaai/jina-clip-v1 open weights, 0.2B 32.4% Image classification huggingface.co Official leaderboard
43 jinaai/jina-embeddings-v5-omni-nano open weights, 1B 29.9% Image classification huggingface.co Official leaderboard
44 Haon-Chen/e5-omni-7B open weights, 8.9B 25.9% Zero-shot classification huggingface.co Official leaderboard
45 Salesforce/blip-itm-base-coco open weights, 0.2B 24% Image classification huggingface.co Official leaderboard
46 mongodb/voyage-multimodal-3 closed weights 22.5% Image classification huggingface.co Official leaderboard
47 Salesforce/blip-itm-large-coco open weights, 0.4B 22.4% Image classification huggingface.co Official leaderboard
48 TIGER-Lab/VLM2Vec-LoRA open weights, 4.2B 22.1% Image classification huggingface.co Official leaderboard
49 TIGER-Lab/VLM2Vec-Full open weights, 4.1B 22% Image classification huggingface.co Official leaderboard
50 Salesforce/blip-itm-large-flickr open weights, 0.4B 21.8% Image classification huggingface.co Official leaderboard
51 Salesforce/blip-image-captioning-large open weights, 0.4B 20.6% Image classification huggingface.co Official leaderboard
52 Salesforce/blip-itm-base-flickr open weights, 0.2B 20.1% Image classification huggingface.co Official leaderboard
53 BidirLM/BidirLM-Omni-2.5B-Embedding open weights, 2.4B 19.7% Zero-shot classification huggingface.co Official leaderboard
54 nyu-visionx/moco-v3-vit-b open weights, 0.1B 18.9% Image classification github.com Official leaderboard
55 nyu-visionx/moco-v3-vit-l open weights, 0.3B 18.6% Image classification github.com Official leaderboard
56 Tevatron/OmniEmbed-v0.1 open weights, 8.9B 17.7% Zero-shot classification huggingface.co Official leaderboard
57 Salesforce/blip-image-captioning-base open weights, 0.2B 17.3% Image classification huggingface.co Official leaderboard
58 Haon-Chen/e5-omni-3B open weights, 4.7B 12.4% Zero-shot classification huggingface.co Official leaderboard
#ModelScoreSettingDateLinkSource
1 EmbeddingGemma 2 39.7% Zero-shot classification Oct 2026 huggingface.co Official leaderboard

Scores over time

All the scores we have, one dot each, so a model tested at several settings appears more than once.

On MTEB’s leaderboard
0% 20% 40% 60% 80% 100% Near the ceiling (85%) EmbeddingGemma 2 0% 20% 40% 60% 80% 100% Near the ceiling (85%) EmbeddingGemma 2

Source: MTEB’s 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 10 checks pass. Every benchmark

Catalog record 2 pass
  • Pass Task access All tasks are public.
  • Pass Confidence Checked against a primary source.
Scores 4 pass 1 note
  • Pass Independent runs Scores from the official leaderboard (58 entries).
  • Pass Models scored 1 model we track, plus 57 other entries on its leaderboard.
  • Pass Newest score The newest is from October 2026, for EmbeddingGemma 2.
  • Note Frontier models None of its models are on the capabilities index.
  • Pass Headroom The best is 39.7%, under 85%, so it still separates models.
Leaderboard import 3 pass
  • Pass Importer We read the board on MTEB: 1 row for models we track and 57 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