PatchCamelyon
Models identify whether a microscope image patch contains metastatic tumor tissue.
MTEB reports accuracy on its PatchCamelyon task.
- Made by
- Bastiaan Veeling and colleagues
- Released
- June 2018
- Kind
- Medicine
- Level
- Expert
- Scored
- Checked automatically
- Answers
- Prediction
- Tasks
- All public
- Measures
- Capability
- Subject
- Clinical medicine
- Language
- English
- Interaction
- Single turn
- Items from
- Real work and use
- Refresh
- One fixed set
- Run by
- Anyone
- In views
- Vision, Biology and medicine
- Last model added
- October 2026
- Scores refreshed
- November 20th, 2026
encord-team/ebind-full tops the leaderboard with 79.5% EmbeddingGemma 2 leads with 69.3%
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 69.3%. This view shows only models we track, each at its best.
- 79.5 encord-team/ebind-full
- 77.3 google/siglip-large-patch16-384
- 76.5 QuanSun/EVA02-CLIP-bigE-14-plus
- 75.5 google/siglip-large-patch16-256
- 75.4 google/siglip-so400m-patch14-384
- 74.7 openai/clip-vit-large-patch14
- 74.7 jinaai/jina-embeddings-v5-omni-nano
- 74.5 facebook/dinov2-small
- 74.4 facebook/dinov2-large
- 74.2 facebook/dinov2-giant
- 74.1 laion/CLIP-ViT-g-14-laion2B-s34B-b88K
- 73.9 google/siglip-base-patch16-256-multilingual
- 73.6 google/siglip-base-patch16-384
- 73.5 laion/CLIP-ViT-L-14-DataComp.XL-s13B-b90K
- 73.5 google/siglip-base-patch16-512
- 73.2 Salesforce/blip2-opt-2.7b
- 73.1 google/siglip-base-patch16-256
- 73.1 nyu-visionx/moco-v3-vit-l
- 73.1 google/siglip-base-patch16-224
- 73.0 laion/CLIP-ViT-H-14-laion2B-s32B-b79K
- 73.0 nyu-visionx/moco-v3-vit-b
- 72.8 jinaai/jina-embeddings-v5-omni-small
- 72.8 QuanSun/EVA02-CLIP-bigE-14
- 72.7 facebook/dinov2-base
- 72.7 BidirLM/BidirLM-Omni-2.5B-Embedding
- 72.4 royokong/e5-v
- 72.3 laion/CLIP-ViT-L-14-laion2B-s32B-b82K
- 71.7 Salesforce/blip2-opt-6.7b-coco
- 71.3 laion/CLIP-ViT-bigG-14-laion2B-39B-b160k
- 71.0 openai/clip-vit-base-patch32
- 70.6 openai/clip-vit-base-patch16
- 70.2 laion/CLIP-ViT-B-16-DataComp.XL-s13B-b90K
- 69.9 laion/CLIP-ViT-B-32-DataComp.XL-s13B-b90K
- 69.7 laion/CLIP-ViT-B-32-laion2B-s34B-b79K
- 69.3
EmbeddingGemma 2
- 69.0 QuanSun/EVA02-CLIP-L-14
- 68.3 kakaobrain/align-base
- 68.3 BAAI/bge-visualized-m3
- 68.2 TIGER-Lab/VLM2Vec-Full
- 68.2 TIGER-Lab/VLM2Vec-LoRA
- 67.8 mongodb/voyage-multimodal-3
- 67.7 Salesforce/blip-itm-large-coco
- 67.7 LCO-Embedding/LCO-Embedding-Omni-7B
- 66.7 Salesforce/blip-image-captioning-large
- 66.1 Salesforce/blip-itm-large-flickr
- 65.7 nomic-ai/nomic-embed-vision-v1.5
- 65.6 QuanSun/EVA02-CLIP-B-16
- 65.3 Tevatron/OmniEmbed-v0.1
- 65.2 BAAI/bge-visualized-base
- 65.0 LCO-Embedding/LCO-Embedding-Omni-3B
- 64.6 Salesforce/blip-itm-base-coco
- 64.1 Haon-Chen/e5-omni-7B
- 64.0 Salesforce/blip-itm-base-flickr
- 63.9 dnotitia/DNA-VL-STEER-2B
- 63.4 Salesforce/blip-image-captioning-base
- 63.0 jinaai/jina-clip-v1
- 62.8 Haon-Chen/e5-omni-3B
- 1 encord-team/ebind-full 79.5%
- 2 google/siglip-large-patch16-384 77.3%
- 3 QuanSun/EVA02-CLIP-bigE-14-plus 76.5%
- 4 google/siglip-large-patch16-256 75.5%
- 5 google/siglip-so400m-patch14-384 75.4%
- 6 openai/clip-vit-large-patch14 74.7%
- 7 jinaai/jina-embeddings-v5-omni-nano 74.7%
- 8 facebook/dinov2-small 74.5%
- 9 facebook/dinov2-large 74.4%
- 10 facebook/dinov2-giant 74.2%
- 11 laion/CLIP-ViT-g-14-laion2B-s34B-b88K 74.1%
- 12 google/siglip-base-patch16-256-multilingual 73.9%
- 13 google/siglip-base-patch16-384 73.6%
- 14 laion/CLIP-ViT-L-14-DataComp.XL-s13B-b90K 73.5%
- 15 google/siglip-base-patch16-512 73.5%
- 16 Salesforce/blip2-opt-2.7b 73.2%
- 17 google/siglip-base-patch16-256 73.1%
- 18 nyu-visionx/moco-v3-vit-l 73.1%
- 19 google/siglip-base-patch16-224 73.1%
- 20 laion/CLIP-ViT-H-14-laion2B-s32B-b79K 73%
- 21 nyu-visionx/moco-v3-vit-b 73%
- 22 jinaai/jina-embeddings-v5-omni-small 72.8%
- 23 QuanSun/EVA02-CLIP-bigE-14 72.8%
- 24 facebook/dinov2-base 72.7%
- 25 BidirLM/BidirLM-Omni-2.5B-Embedding 72.7%
- 26 royokong/e5-v 72.4%
- 27 laion/CLIP-ViT-L-14-laion2B-s32B-b82K 72.3%
- 28 Salesforce/blip2-opt-6.7b-coco 71.7%
- 29 laion/CLIP-ViT-bigG-14-laion2B-39B-b160k 71.3%
- 30 openai/clip-vit-base-patch32 71%
- 31 openai/clip-vit-base-patch16 70.6%
- 32 laion/CLIP-ViT-B-16-DataComp.XL-s13B-b90K 70.2%
- 33 laion/CLIP-ViT-B-32-DataComp.XL-s13B-b90K 69.9%
- 34 laion/CLIP-ViT-B-32-laion2B-s34B-b79K 69.7%
- 35
EmbeddingGemma 2 69.3%
- 36 QuanSun/EVA02-CLIP-L-14 69%
- 37 kakaobrain/align-base 68.3%
- 38 BAAI/bge-visualized-m3 68.3%
- 39 TIGER-Lab/VLM2Vec-Full 68.2%
- 40 TIGER-Lab/VLM2Vec-LoRA 68.2%
- 41 mongodb/voyage-multimodal-3 67.8%
- 42 Salesforce/blip-itm-large-coco 67.7%
- 43 LCO-Embedding/LCO-Embedding-Omni-7B 67.7%
- 44 Salesforce/blip-image-captioning-large 66.7%
- 45 Salesforce/blip-itm-large-flickr 66.1%
- 46 nomic-ai/nomic-embed-vision-v1.5 65.7%
- 47 QuanSun/EVA02-CLIP-B-16 65.6%
- 48 Tevatron/OmniEmbed-v0.1 65.3%
- 49 BAAI/bge-visualized-base 65.2%
- 50 LCO-Embedding/LCO-Embedding-Omni-3B 65%
- 51 Salesforce/blip-itm-base-coco 64.6%
- 52 Haon-Chen/e5-omni-7B 64.1%
- 53 Salesforce/blip-itm-base-flickr 64%
- 54 dnotitia/DNA-VL-STEER-2B 63.9%
- 55 Salesforce/blip-image-captioning-base 63.4%
- 56 jinaai/jina-clip-v1 63%
- 57 Haon-Chen/e5-omni-3B 62.8%
Source: MTEB’s leaderboard
Source: MTEB’s leaderboard
| # | Entry | Score | Setting | Date | Link | Source |
|---|---|---|---|---|---|---|
| 1 | encord-team/ebind-full open weights, 1.8B | 79.5% | huggingface.co | Official leaderboard | ||
| 2 | google/siglip-large-patch16-384 open weights, 0.7B | 77.3% | huggingface.co | Official leaderboard | ||
| 3 | QuanSun/EVA02-CLIP-bigE-14-plus open weights, 5B | 76.5% | huggingface.co | Official leaderboard | ||
| 4 | google/siglip-large-patch16-256 open weights, 0.7B | 75.5% | huggingface.co | Official leaderboard | ||
| 5 | google/siglip-so400m-patch14-384 open weights, 0.9B | 75.4% | huggingface.co | Official leaderboard | ||
| 6 | openai/clip-vit-large-patch14 open weights, 0.4B | 74.7% | huggingface.co | Official leaderboard | ||
| 7 | jinaai/jina-embeddings-v5-omni-nano open weights, 1B | 74.7% | huggingface.co | Official leaderboard | ||
| 8 | facebook/dinov2-small open weights, 0B | 74.5% | huggingface.co | Official leaderboard | ||
| 9 | facebook/dinov2-large open weights, 0.3B | 74.4% | huggingface.co | Official leaderboard | ||
| 10 | facebook/dinov2-giant open weights, 1.1B | 74.2% | huggingface.co | Official leaderboard | ||
| 11 | laion/CLIP-ViT-g-14-laion2B-s34B-b88K open weights, 1.4B | 74.1% | huggingface.co | Official leaderboard | ||
| 12 | google/siglip-base-patch16-256-multilingual open weights, 0.4B | 73.9% | huggingface.co | Official leaderboard | ||
| 13 | google/siglip-base-patch16-384 open weights, 0.2B | 73.6% | huggingface.co | Official leaderboard | ||
| 14 | laion/CLIP-ViT-L-14-DataComp.XL-s13B-b90K open weights, 0.4B | 73.5% | huggingface.co | Official leaderboard | ||
| 15 | google/siglip-base-patch16-512 open weights, 0.2B | 73.5% | huggingface.co | Official leaderboard | ||
| 16 | Salesforce/blip2-opt-2.7b open weights, 3.7B | 73.2% | huggingface.co | Official leaderboard | ||
| 17 | google/siglip-base-patch16-256 open weights, 0.2B | 73.1% | huggingface.co | Official leaderboard | ||
| 18 | nyu-visionx/moco-v3-vit-l open weights, 0.3B | 73.1% | github.com | Official leaderboard | ||
| 19 | google/siglip-base-patch16-224 open weights, 0.2B | 73.1% | huggingface.co | Official leaderboard | ||
| 20 | laion/CLIP-ViT-H-14-laion2B-s32B-b79K open weights, 1B | 73% | huggingface.co | Official leaderboard | ||
| 21 | nyu-visionx/moco-v3-vit-b open weights, 0.1B | 73% | github.com | Official leaderboard | ||
| 22 | jinaai/jina-embeddings-v5-omni-small open weights, 1.6B | 72.8% | huggingface.co | Official leaderboard | ||
| 23 | QuanSun/EVA02-CLIP-bigE-14 open weights, 4.7B | 72.8% | huggingface.co | Official leaderboard | ||
| 24 | facebook/dinov2-base open weights, 0.1B | 72.7% | huggingface.co | Official leaderboard | ||
| 25 | BidirLM/BidirLM-Omni-2.5B-Embedding open weights, 2.4B | 72.7% | huggingface.co | Official leaderboard | ||
| 26 | royokong/e5-v open weights, 8.4B | 72.4% | huggingface.co | Official leaderboard | ||
| 27 | laion/CLIP-ViT-L-14-laion2B-s32B-b82K open weights, 0.4B | 72.3% | huggingface.co | Official leaderboard | ||
| 28 | Salesforce/blip2-opt-6.7b-coco open weights, 7.8B | 71.7% | huggingface.co | Official leaderboard | ||
| 29 | laion/CLIP-ViT-bigG-14-laion2B-39B-b160k open weights, 2.5B | 71.3% | huggingface.co | Official leaderboard | ||
| 30 | openai/clip-vit-base-patch32 open weights, 0.2B | 71% | huggingface.co | Official leaderboard | ||
| 31 | openai/clip-vit-base-patch16 open weights, 0.2B | 70.6% | huggingface.co | Official leaderboard | ||
| 32 | laion/CLIP-ViT-B-16-DataComp.XL-s13B-b90K open weights, 0.2B | 70.2% | huggingface.co | Official leaderboard | ||
| 33 | laion/CLIP-ViT-B-32-DataComp.XL-s13B-b90K open weights, 0.2B | 69.9% | huggingface.co | Official leaderboard | ||
| 34 | laion/CLIP-ViT-B-32-laion2B-s34B-b79K open weights, 0.2B | 69.7% | huggingface.co | Official leaderboard | ||
| 35 | | 69.3% | Oct 2026 | huggingface.co | Official leaderboard | |
| 36 | QuanSun/EVA02-CLIP-L-14 open weights, 0.4B | 69% | huggingface.co | Official leaderboard | ||
| 37 | kakaobrain/align-base open weights, 0.2B | 68.3% | huggingface.co | Official leaderboard | ||
| 38 | BAAI/bge-visualized-m3 open weights, 0.9B | 68.3% | huggingface.co | Official leaderboard | ||
| 39 | TIGER-Lab/VLM2Vec-Full open weights, 4.1B | 68.2% | huggingface.co | Official leaderboard | ||
| 40 | TIGER-Lab/VLM2Vec-LoRA open weights, 4.2B | 68.2% | huggingface.co | Official leaderboard | ||
| 41 | mongodb/voyage-multimodal-3 closed weights | 67.8% | huggingface.co | Official leaderboard | ||
| 42 | Salesforce/blip-itm-large-coco open weights, 0.4B | 67.7% | huggingface.co | Official leaderboard | ||
| 43 | LCO-Embedding/LCO-Embedding-Omni-7B open weights, 8.9B | 67.7% | huggingface.co | Official leaderboard | ||
| 44 | Salesforce/blip-image-captioning-large open weights, 0.4B | 66.7% | huggingface.co | Official leaderboard | ||
| 45 | Salesforce/blip-itm-large-flickr open weights, 0.4B | 66.1% | huggingface.co | Official leaderboard | ||
| 46 | nomic-ai/nomic-embed-vision-v1.5 open weights, 0.1B | 65.7% | huggingface.co | Official leaderboard | ||
| 47 | QuanSun/EVA02-CLIP-B-16 open weights, 0.1B | 65.6% | huggingface.co | Official leaderboard | ||
| 48 | Tevatron/OmniEmbed-v0.1 open weights, 8.9B | 65.3% | huggingface.co | Official leaderboard | ||
| 49 | BAAI/bge-visualized-base open weights, 0.2B | 65.2% | huggingface.co | Official leaderboard | ||
| 50 | LCO-Embedding/LCO-Embedding-Omni-3B open weights, 4.7B | 65% | huggingface.co | Official leaderboard | ||
| 51 | Salesforce/blip-itm-base-coco open weights, 0.2B | 64.6% | huggingface.co | Official leaderboard | ||
| 52 | Haon-Chen/e5-omni-7B open weights, 8.9B | 64.1% | huggingface.co | Official leaderboard | ||
| 53 | Salesforce/blip-itm-base-flickr open weights, 0.2B | 64% | huggingface.co | Official leaderboard | ||
| 54 | dnotitia/DNA-VL-STEER-2B open weights, 2.1B | 63.9% | huggingface.co | Official leaderboard | ||
| 55 | Salesforce/blip-image-captioning-base open weights, 0.2B | 63.4% | huggingface.co | Official leaderboard | ||
| 56 | jinaai/jina-clip-v1 open weights, 0.2B | 63% | huggingface.co | Official leaderboard | ||
| 57 | Haon-Chen/e5-omni-3B open weights, 4.7B | 62.8% | huggingface.co | Official leaderboard |
| # | Model | Score | Setting | Date | Link | Source |
|---|---|---|---|---|---|---|
| 1 | | 69.3% | 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.
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 Law, finance and medicine
- MedXpertQA-MM In use Models answer expert-level medical exam questions that combine images with clinical records and test results. Scoring is accuracy.
Qwen 3.8 Max 80.4%
- HealthBench Professional In use 525 physician-authored tasks from real clinician chats (care consult, documentation, research) with physician-adjudicated rubrics. Claude Sonnet 5.5 69.2%
- HealthBench In use 5,000 realistic health conversations graded against physician-written rubrics. A length-adjusted variant penalizes verbosity. GPT-5 67.2%
- FinSearchComp In use Agents search the web to fetch live financial data, look up history, and run multi-step investigations, across 635 expert-written questions. Scored by answer accuracy.
Ling-3.1-flash 81.7%
- KorMedMCQA In use Models answer 7,469 multiple-choice questions from Korean licensing exams for doctors, nurses, pharmacists and dentists, held from 2012 to 2024. o1-preview 92.7%
- Harvey LAB-AA v1.1 In use Artificial Analysis's run of Harvey's Legal Agent Benchmark on Harvey's newer private set of 120 tasks across 24 practice areas. A deliverable with a material hallucination scores zero, and the headline is the share of judges who find every rubric criterion passed. GPT-6.2 Astra 15%
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 (57 entries).
- Pass Models scored 1 model we track, plus 56 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 69.3%, 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 56 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.