Embedding Model Rankings

Best Embedding Models

Text and multimodal embedding models ranked by the MTEB leaderboard: retrieval, classification, clustering, and STS.

Benchmark data last synced May 7, 2026

27 models


AA74.7	Qwen3-Embedding-0.6B Qwen/Alibaba	65 Average	1024	33K	0.596B
AA72.7	multilingual-e5-large-instruct intfloat (Microsoft Research)	62 Average	1024	514	0.56B
AA72.5	harrier-oss-v1-0.6b Microsoft	69 Average	1024	33K	0.596B
AA72.2	Qwen3-Embedding-4B Qwen/Alibaba	70 Average	2560	33K	4B
AA71.9	Qwen3-Embedding-8B Qwen/Alibaba	71 Average	4096	33K	7.6B
AA71.1	jina-embeddings-v5-text-small Jina AI	68 Average	1024	33K	0.596B
AA70.3	harrier-oss-v1-270m Microsoft	66 Average	640	33K	0.268B
BB69.3	jina-embeddings-v5-text-nano Jina AI	67 Average	768	8K	0.212B
BB63.5	gte-Qwen2-7B-instruct Alibaba	64 Average	3584	33K	7.1B
BB61.4	llama-embed-nemotron-8b NVIDIA	69 Average	4096	33K	7.5B
BB61.0	Octen-Embedding-8B Octen AI	69 Average	4096	33K	7.6B
BB60.2	e5-mistral-7b-instruct intfloat (Microsoft Research)	61 Average	4096	33K	7.1B
BB59.0	harrier-oss-v1-27b Microsoft	74 Average	5376	131K	27B
BB57.2	KaLM-Embedding-Gemma3-12B-2511 Tencent	72 Average	3840	33K	11.8B
BB55.4	SFR-Embedding-Mistral Salesforce	63 Average	4096	33K	7.1B

Rows per page

Page 1 of 2

Need Help Choosing the Right Model?

Our team can help you find the right AI model for your specific hardware and use case.