Qwen-Turbo, based on Qwen2.5, is a 1M context model developed by Alibaba Cloud, offering fast speed and low cost, making it suitable for simple tasks. It extends the context length to 1 million tokens, equivalent to approximately 1 million English words or 1.5 million Chinese characters. This model achieves exceptional performance in long-text tasks, scoring 93.1 on the RULER benchmark, and utilizes sparse attention mechanisms to reduce inference time significantly.

Qwen-Turbo maintains strong performance in short-text tasks, similar to GPT-4o-mini, and supports multiple languages and multimodal capabilities, including natural language understanding, text generation, vision understanding, and audio understanding.