Embedding Models¶

Models that convert text, images, or audio into dense vectors for semantic search, similarity, and retrieval-augmented generation (RAG).

Text Embeddings¶

Model	Dims	Size	Browser?	License	Link
all-MiniLM-L6-v2	384	23 MB	Yes	Apache-2.0	HF
bge-small-en-v1.5	384	33 MB	Yes	MIT	HF
gte-small	384	67 MB	Yes	MIT	HF
multilingual-e5-small	384	118 MB	Yes	MIT	HF

Model	Dims	Size	Context	License	Link
bge-large-en-v1.5	1024	335 MB	512	MIT	HF
e5-large-v2	1024	330 MB	512	MIT	HF
nomic-embed-text-v1.5	768	274 MB	8192	Apache-2.0	HF
jina-embeddings-v2-base	768	137 MB	8192	Apache-2.0	HF
mxbai-embed-large	1024	335 MB	512	Apache-2.0	HF

Model	Dims	Size	License	Link
CLIP ViT-B/32	512	150 MB	MIT	HF
CLIP ViT-L/14	768	890 MB	MIT	HF
SigLIP	768	400 MB	Apache-2.0	HF
DINOv2	768	86 MB (small)	Apache-2.0	HF

Model	Modalities	Dims	License	Link
CLIP	Text + Image	512/768	MIT	HF
ImageBind	Text + Image + Audio + Video + IMU + Thermal	1024	CC-BY-NC-SA	HF
ONE-PEACE	Text + Image + Audio	1536	Apache-2.0	HF

Database	Browser?	License	Link
vectra (in-memory)	Yes	MIT	github.com/Stevenic/vectra
Voy (WASM)	Yes	Apache-2.0	github.com/tantaraio/voy
Chroma	No (Python)	Apache-2.0	trychroma.com
Qdrant	No (server)	Apache-2.0	qdrant.tech