Qwen3-Embedding-8B is a 7.6 billion parameter text embedding model developed by Qwen, part of the Qwen3 Embedding series. It is specifically designed for text embedding and ranking tasks, inheriting strong multilingual capabilities and long-text understanding from its foundational Qwen3 models. This model achieves state-of-the-art performance, ranking #1 on the MTEB multilingual leaderboard with a score of 70.58, and supports over 100 languages with a 32K context length and user-defined embedding dimensions up to 4096.
Loading preview...
Qwen3-Embedding-8B: A High-Performance Multilingual Text Embedding Model
Qwen3-Embedding-8B is the largest model in the Qwen3 Embedding series, a proprietary line of models from Qwen specifically engineered for advanced text embedding and ranking. Built upon the robust Qwen3 foundational models, this 7.6 billion parameter model excels in various natural language processing tasks, including text retrieval, code retrieval, classification, clustering, and bitext mining.
Key Capabilities & Features
- Exceptional Multilingual Performance: Achieves a leading score of 70.58 on the MTEB multilingual leaderboard (as of June 5, 2025), supporting over 100 languages, including programming languages.
- Long Context Understanding: Features a substantial context length of 32,768 tokens, enabling comprehensive analysis of lengthy texts.
- Flexible Embedding Dimensions: Supports user-defined output dimensions ranging from 32 to 4096, offering adaptability for diverse application requirements.
- Instruction-Aware: Allows for user-defined instructions to optimize performance for specific tasks, languages, or scenarios, with a recommended 1-5% performance improvement when used.
- Comprehensive Series: Part of a broader series that includes both embedding and reranking models in various sizes (0.6B, 4B, 8B), designed for seamless integration.
When to Use This Model
Qwen3-Embedding-8B is particularly well-suited for applications requiring highly accurate and versatile text embeddings across multiple languages. Its top-tier performance in retrieval, classification, and clustering tasks makes it ideal for building sophisticated search engines, recommendation systems, and cross-lingual information retrieval systems. The model's instruction-aware capability further enhances its utility for fine-tuning performance on specific, nuanced tasks.