Qwen3-Reranker 8B

Qwen3-Reranker 8B is the largest text reranking model in Alibaba's Qwen3 embedding series, released alongside the 0.6B and 4B rerankers and the Qwen3-Embedding models. It scores query-document pairs for retrieval-augmented generation pipelines, with a 32K sequence length that accommodates long documents. The series is instruction-aware, and the Qwen team reports that tailored instructions typically lift downstream task quality by one to five percent.

The models are released under Apache 2.0 on both Hugging Face and ModelScope, and community GGUF quantizations enable CPU and llama.cpp deployment. Release date is era-level (June 2025) matching the Qwen3 embedding series launch window.