Alibaba Open Source Qwen3 New Model Performance Surpasses Google Gemini-Embedding

In the early hours of this morning, Alibaba open-sourced two new models of the Qwen3 series, Qwen3-Embedding and Qwen3-Reranker. These two models are specially designed for text representation, retrieval and sorting tasks, and are trained based on the Qwen3 basic model, which fully inherits the advantages of Qwen3 in multilingual text understanding and supports 119 languages. According to the test data, Qwen3 Embedding performs very well in the multilingual text representation benchmark. Among them, the 8B parameter ranked first with a high score of 70.58, surpassing many commercial API services, such as Google’s Gemini-Embedding.