Alibaba Cloud Launches Qwen3-VL-Flash: Upgraded Long Text Understanding and Spatial Awareness Capabilities
Alibaba Cloud's model service has introduced a new visual language model, Qwen3-VL-Flash, supporting up to 256K context length, with enhanced performance in image and video understanding and multimodal tasks, while offering faster response times and lower costs.

Alibaba Cloud's model service has launched Qwen3-VL-Flash, a visual language model that combines both reasoning and non-reasoning modes.
Officially, its performance surpasses the open-source Qwen3-VL-30B-A3B and Qwen2.5-72B, with faster response times, stronger capabilities, and lower costs.
Key features include support for up to 256K tokens in very long contexts, suitable for handling long videos and documents; enhanced image and video understanding capabilities, supporting 2D/3D localization and spatial perception; improved OCR, multilingual recognition, agent control, and practical application capabilities; as well as significant advancements in security perception and real-world visual intelligence.
Some users have reported issues with the mobile experience. The image below shows a page prompt that mobile access is not supported, recommending the use of a computer.
![This is a screenshot showing that the Alibaba Cloud console is not usable on mobile devices. The top of the screen displays the current time as 23:47, with battery level and network signal information in the upper right corner. The page title is 'Alibaba Cloud Bailian,' with an illustration of a laptop below it, the screen of which shows an icon similar to a folder. Below the illustration, there is a text prompt: 'The current page does not currently support the mobile experience. Please proceed to the computer for the experience.' Additionally, there is a URL link 'https://bailian.console.aliyun.com.' The background color of the entire page is white, with the main text in black.]
发布时间: 2025-10-16 23:13