l

CLIP ViT B 32 DataComp.XL S13B B90K

Apr 2023
lCLIP ViT B 32 DataComp.XL S13B B90K
Compare in
on
self-host onlyprice
11.7MIEB lite
Apr 2023released
0K tokensMax input

Ranking?

Full ranking →
1EVA02 CLIP B 1643.22Siglip So400m Patch14 38441.63Siglip Large Patch16 38441.44CLIP ViT BigG 14 Laion2B 39B B160k41.25Dinov2 Giant40.8
···
22Dinov2 Small34.723CLIP ViT B 32 DataComp.XL S13B B90K34.224CLIP ViT B 32 Laion2B S34B B79K33.9

Strong at

C
Embeddings & Retrieval#85 of 89

Alternatives?

ModelScoreCostPriceWhat changes
BetterJina Embeddings v5 Omni SmallunstatedOpen+9%—$0.05Memory (4-bit): 0.1 GB → 1 GBQOpenEVA02 CLIP B 16unstated+26%——Memory (4-bit): 0.1 GB → 3 GBBetterSiglip So400m Patch14 384Open+22%——Memory (4-bit): 0.1 GB → 0.5 GBBetterSiglip Large Patch16 384Open+21%——Memory (4-bit): 0.1 GB → 0.4 GBlBetterCLIP ViT BigG 14 Laion2B 39B B160kOpen+21%——Memory (4-bit): 0.1 GB → 1.5 GB

Run it yourself

estimate
Fits on✓ 8 GB (phone, laptop) · full quality✓ 16 GB · full quality✓ 24 GB GPU · full quality✓ 48 GB GPU · full quality✓ 80 GB GPU · full quality
LicenceMITfree for commercial use

Benchmarks

in Embeddings & Retrieval
MIEB English · Mean over tasks · multilingual (13 languages)49.9#27
MIEB English · Mean over task types · multilingual (13 languages)49.9#27
MIEB English · Document Understanding · English12.4#76
MIEB English · Zero-shot Classification · English58.7#19
MIEB English · Any-to-any Retrieval · English32.2#24
MIEB English · Visual semantic similarity (English) · multilingual (13 languages)56.5#35
MIEB English · Image Classification · English75.8#20
MIEB English · Compositionality · English39.7#22
MIEB English · Vision Centric QA · English46.1#27
MIEB English · Image Clustering · English77.4#21
MIEB Img · Mean over tasks · multilingual (13 languages)53.9#28
MIEB Img · Mean over task types · multilingual (13 languages)53.9#28
MIEB Img · Image Classification · English75.8#20
MIEB Img · Any-to-any Retrieval · English34.2#23
MIEB Img · Visual semantic similarity (English) · English54.6#35
MIEB Img · Visual semantic similarity (multilingual) · multilingual (13 languages)27.5#40
MIEB Img · Image Clustering · English77.4#21
MIEB lite · Mean over tasks · multilingual (41 languages)41.5#36
MIEB lite · Mean over task types · multilingual (41 languages)41.5#36
MIEB lite · Vision Centric QA · English47.3#36
MIEB lite · Any-to-any Retrieval · English26.3#31
MIEB lite · Document Understanding · English11.7#85
MIEB lite · Zero-shot Classification · English58.2#23
MIEB lite · Compositionality · English33.1#26
MIEB lite · Image Classification · English71#23
MIEB lite · Any-to-any Multilingual Retrieval · multilingual (40 languages)18.7#35
MIEB lite · Image Clustering · English70.3#27
MIEB lite · Visual semantic similarity (English) · English54.9#44
MIEB lite · Visual semantic similarity (multilingual) · multilingual (13 languages)23.6#46
MIEB multilingual · Mean over tasks · multilingual (41 languages)44.3#30
MIEB multilingual · Mean over task types · multilingual (41 languages)44.3#30
MIEB multilingual · Zero-shot Classification · English58.7#19
MIEB multilingual · Image Classification · English75.8#20
MIEB multilingual · Document Understanding · English12.4#76
MIEB multilingual · Any-to-any Retrieval · English32.2#24
MIEB multilingual · Compositionality · English39.7#22
MIEB multilingual · Vision Centric QA · English46.1#27
MIEB multilingual · Visual semantic similarity (English) · multilingual (13 languages)56.5#35
MIEB multilingual · Visual semantic similarity (multilingual) · multilingual (13 languages)23.6#46
MIEB multilingual · Any-to-any Multilingual Retrieval · multilingual (40 languages)20.1#25
MIEB multilingual · Image Clustering · English77.4#21

FAQ

How much does CLIP ViT B 32 DataComp.XL S13B B90K cost?›

There is no per-unit API price: self-host only.

Is CLIP ViT B 32 DataComp.XL S13B B90K free to use?›

There is no free API. The weights are open, so you can run it yourself.

How good is CLIP ViT B 32 DataComp.XL S13B B90K, and what is it best at?›

#85 of 89 in Multimodal.

Is CLIP ViT B 32 DataComp.XL S13B B90K better than Dinov2 Small?›

On MIEB Img · Any-to-any Retrieval · English, CLIP ViT B 32 DataComp.XL S13B B90K scores 34.2 vs 34.7 — about the same.

When was CLIP ViT B 32 DataComp.XL S13B B90K released? Is there a newer version?›

Released 2023-04-26. It is the current version in its line.

Is CLIP ViT B 32 DataComp.XL S13B B90K open source?›

Yes, the weights are open (MIT licence).

Can I run CLIP ViT B 32 DataComp.XL S13B B90K locally?›

The weights are open and can be self-hosted; hardware needs are not confirmed yet.

What is CLIP ViT B 32 DataComp.XL S13B B90K?›

CLIP ViT B 32 DataComp.XL S13B B90K is an open-weights embeddings & retrieval model from laion, released Apr 2023.

What are the best alternatives to CLIP ViT B 32 DataComp.XL S13B B90K?›

Jina Embeddings v5 Omni Small (better), EVA02 CLIP B 16 (open weights).