Siglip Large Patch16 256

Jan 2024
Siglip Large Patch16 256
Compare in
on
self-host onlyprice
31.9MIEB lite
Jan 2024released
1,024Dimensions
0K tokensMax input

Ranking?

Full ranking →
1Voyage Multimodal 381.82Jina Embeddings v5 Omni Small81.73DNA VL STEER 2B80.74E5 V79.35Jina Embeddings v5 Omni Nano75.9
···
16Siglip Base Patch16 51267.717Siglip Large Patch16 25667.418Siglip Base Patch16 38467.1

Strong at

C
Embeddings & Retrieval#64 of 89

Alternatives?

ModelScoreCostPriceWhat changes
BetterJina Embeddings v5 Omni NanounstatedOpen+13%—$0.02Memory (4-bit): 0.4 GB → 0.6 GBOpenJina Embeddings v5 Omni Smallunstated+21%—$0.05Memory (4-bit): 0.4 GB → 1 GBVBetterVoyage Multimodal 3+21%—$0.12dBetterDNA VL STEER 2BOpen+20%——Memory (4-bit): 0.4 GB → 1.3 GBrBetterE5 VOpen+18%——Memory (4-bit): 0.4 GB → 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
LicenceApache-2.0free for commercial use

Benchmarks

in Embeddings & Retrieval
MIEB English · Mean over tasks · multilingual (13 languages)57.9#13
MIEB English · Mean over task types · multilingual (13 languages)57.9#13
MIEB English · Document Understanding · English39.4#56
MIEB English · Zero-shot Classification · English67.7#7
MIEB English · Any-to-any Retrieval · English38.8#9
MIEB English · Visual semantic similarity (English) · multilingual (13 languages)67.4#17
MIEB English · Image Classification · English82.5#7
MIEB English · Compositionality · English40.7#16
MIEB English · Vision Centric QA · English44.9#34
MIEB English · Image Clustering · English82.1#12
MIEB Img · Mean over tasks · multilingual (13 languages)62.6#8
MIEB Img · Mean over task types · multilingual (13 languages)62.6#8
MIEB Img · Image Classification · English82.5#7
MIEB Img · Any-to-any Retrieval · English39.6#9
MIEB Img · Visual semantic similarity (English) · English66.1#17
MIEB Img · Visual semantic similarity (multilingual) · multilingual (13 languages)42.8#17
MIEB Img · Image Clustering · English82.1#12
MIEB lite · Mean over tasks · multilingual (41 languages)51.4#15
MIEB lite · Mean over task types · multilingual (41 languages)51.4#15
MIEB lite · Vision Centric QA · English46.5#42
MIEB lite · Any-to-any Retrieval · English31#18
MIEB lite · Document Understanding · English31.9#64
MIEB lite · Zero-shot Classification · English70.3#6
MIEB lite · Compositionality · English33.4#25
MIEB lite · Image Classification · English76.4#7
MIEB lite · Any-to-any Multilingual Retrieval · multilingual (40 languages)42.6#14
MIEB lite · Image Clustering · English76.5#17
MIEB lite · Visual semantic similarity (English) · English67.6#28
MIEB lite · Visual semantic similarity (multilingual) · multilingual (13 languages)38.1#23
MIEB multilingual · Mean over tasks · multilingual (41 languages)55.2#7
MIEB multilingual · Mean over task types · multilingual (41 languages)55.2#7
MIEB multilingual · Zero-shot Classification · English67.7#7
MIEB multilingual · Image Classification · English82.5#7
MIEB multilingual · Document Understanding · English39.4#56
MIEB multilingual · Any-to-any Retrieval · English38.8#9
MIEB multilingual · Compositionality · English40.7#16
MIEB multilingual · Vision Centric QA · English44.9#34
MIEB multilingual · Visual semantic similarity (English) · multilingual (13 languages)67.4#17
MIEB multilingual · Visual semantic similarity (multilingual) · multilingual (13 languages)38.1#23
MIEB multilingual · Any-to-any Multilingual Retrieval · multilingual (40 languages)49.8#7
MIEB multilingual · Image Clustering · English82.1#12

FAQ

How much does Siglip Large Patch16 256 cost?›

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

Is Siglip Large Patch16 256 free to use?›

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

How good is Siglip Large Patch16 256, and what is it best at?›

#64 of 89 in Multimodal.

Is Siglip Large Patch16 256 better than Siglip Base Patch16 512?›

On MIEB English · Visual semantic similarity (English) · multilingual (13 languages), Siglip Large Patch16 256 scores 67.4 vs 67.7 — about the same.

When was Siglip Large Patch16 256 released? Is there a newer version?›

Released 2024-01-08. It is the current version in its line.

Is Siglip Large Patch16 256 open source?›

Yes, the weights are open (Apache-2.0 licence).

Can I run Siglip Large Patch16 256 locally?›

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

What is Siglip Large Patch16 256?›

Siglip Large Patch16 256 is an open-weights embeddings & retrieval model from Google, released Jan 2024.

What are the best alternatives to Siglip Large Patch16 256?›

Jina Embeddings v5 Omni Nano (better), Jina Embeddings v5 Omni Small (open weights).