Alibaba’s Qwen Team has released a new open source AI image generator called Qwen-Image. According to its developers, the model specializes in creating images with accurately rendered text in both English and Chinese, a common challenge for many AI systems. Writing for VentureBeat, journalist Carl Franzen notes this capability allows for creating content like posters, presentation slides, or storefront scenes with clear text.
The model is available under the Apache 2.0 license, which permits free commercial use and modification. This makes it an attractive alternative to subscription-based services like Midjourney. However, Franzen’s initial tests revealed that the model’s text accuracy was not noticeably better than its proprietary rivals.
For businesses, the open source nature is a significant advantage. A key drawback is that Alibaba does not disclose the model’s training data and, unlike competitors such as Adobe or OpenAI, provides no legal indemnification against potential copyright claims.
In official benchmarks, Qwen-Image performs strongly, matching or surpassing other models, particularly in rendering Chinese text. On the public evaluation platform AI Arena, it is ranked as the top open source model based on user comparisons. Qwen-Image is accessible on platforms like Hugging Face and GitHub.