Qwen3-Reranker-0.6B

Qwen3-Reranker-0.6B is a reranking model: it does not search, it re-scores. A search engine returns a rough list of candidate documents, and the reranker reads each query-document pair in full and puts the list in a better order. Alibaba released it on 5 June 2025 alongside the Qwen 3 embedding models, in three sizes meant to be mixed and matched with the embedding side. At 0.6 billion parameters across 28 layers it is the cheapest member of the family to serve, and the only one that runs comfortably without a data-centre accelerator. It pays for that with the weakest instruction-following score in the set: 5.41 on FollowIR against 14.84 for the 4B model, meaning it is the least able of the three to obey a written ranking instruction rather than simply matching topic. The model is built on the Qwen3-0.6B-Base checkpoint, handles a 32k-token context and over 100 languages including programming languages, and accepts a written instruction describing what "relevant" means for the task at hand — the maker measures a 1 to 5 percent gain from using one, and advises writing it in English even for other languages, because the training instructions were English. Weights are Apache 2.0, commercial use included; the repository records about 1.25M downloads in the past thirty days.

▸Videos

No videos yet.

▸Reviews