Alibaba has launched its next-generation image generation and editing model, Qwen-Image-2.0 that supports ultra-long text input of up to 1K tokens and 2K high-resolution outputs, enabling the accurate rendering of complex instructions.

As the image generation foundation model for the Qwen large model, Qwen-Image-2.0 integrates both image generation and editing. It scored 1029 points in the AI Arena text-to-image evaluation, ranking third globally.

Currently, the API for Qwen-Image-2.0 is available for invitation testing on Alibaba Cloud's BaiLian platform, and developers can also experience the new model for free via Qwen Chat.