Embeddings and rerankers · Alibaba Qwen

Qwen3 Embedding 8B

Hugging Face: Qwen/Qwen3-Embedding-8B · checked 2026-10-13

Top open embedding model on MMTEB at release; 32K context, 4,096 dimensions.

Datasheet

Maker	Alibaba Qwen
Parameters	7.57 BVerifiedsafetensors.total from the HF API Source
Licence	Apache-2.0Verified
Access	Public download
Task (HF pipeline)	feature-extraction
First published on HF	2025-06-03
Last updated on HF	2025-07-07
Weights in memory	15 GB at 16-bit · 4.6 GB at ~4.85 bitsEstimateParameters × bits ÷ 8. Excludes context cache and runtime overhead; not every model has a 4-bit build.

Benchmark

70.58

MMTEB (multilingual) mean task score; rerankers: MTEB-R · #1 of 4 scored in this category

MTEB (Multilingual), mean (task). Source

Other published results

MTEB (Eng v2), mean (task): 75.22Sourced Source

See the full embeddings and rerankers ranking

Used in

The description summarises the model card at https://huggingface.co/Qwen/Qwen3-Embedding-8B. Parameters and licence are read from the Hugging Face API; the weights-in-memory line is our arithmetic, not a measurement.