Alibaba's open-weight Qwen-Image-2.1 claims to beat closed models in image generation...

Alibaba's open-weight Qwen-Image-2.1 claims to beat closed models in image generation with just 7 billion parameters

Alibaba's open-weight Qwen-Image-2.1 claims to beat closed models in image generation with just 7 billion parameters

Alibaba's Qwen AI team has released Qwen-Image-2.1, an open-weight model for image generation and editing. Its visual generation component has just 7 billion parameters yet beats most closed models on Qwen's own benchmark, the team claims, though independent benchmarks are still pending. It runs on capable consumer GPUs like a 3090.

Qwen says this group portrait was assembled from individual photos of each person. | Image: Alibaba / Qwen

The model natively generates and edits transparent images (RGBA), letting users isolate objects or change text on transparent layers. It handles up to ten reference images at once for group portraits, virtual try-ons, or room design, while circles, masks, or painted marks guide local edits.Qwen saysarchitecture changes and KV cache reuse speed up inference, especially with multiple reference images.

Qwen-Image-2.1 is available onHugging Face,GitHub, andModel Scope, with aHugging Face demo. Itsresearch licensebars commercial use, so business users must apply to Qwen for a separate license. Ad Ad

AI News Without the Hype – Curated by Humans

Subscribe to THE DECODER for ad-free reading, a weekly AI newsletter, our exclusive "AI Radar" frontier report six times a year, full archive access, and access to our comment section.