Qwen released a prompt rewriting model for Qwen-Image-2.1
Qwen released Qwen-Image-2.1-PE-T2I, a fine-tuned Qwen3.5-VL 9B model that rewrites brief image requests into detailed English prompts and recommends aspect ratios for Qwen-Image-2.1. It is available on Hugging Face under the Qwen Research License Agreement.
China context
- Original name
- 通义千问图像2.1提示词重写模型
- Outside China
- Open weights · huggingface.co
- Claims
- Company-reported; not yet independently evaluated
- For builders
- Developers outside China can access the model weights on Hugging Face and integrate prompt rewriting into their image generation workflows, subject to the Qwen Research License Agreement.
- For investors
- The release of a prompt rewriting model indicates Qwen's focus on improving usability of its image generation models, which may increase adoption among developers and potentially drive demand for related services.
Qwen has open-sourced Qwen-Image-2.1-PE-T2I, a text-to-image prompt rewriting model designed to work with Qwen-Image-2.1. The model is a fine-tuned Qwen3.5-VL 9B that converts brief image requests in any language into detailed English prompts and suggests an aspect ratio. It is available on Hugging Face under the Qwen Research License Agreement.
The model is a fine-tuned Qwen3.5-VL 9B that outputs a JSON object containing a rewritten prompt and a recommended aspect ratio after a reasoning block. It is designed to improve prompt quality for Qwen-Image-2.1, which uses a 7B parameter visual generation component with 32 Single-Stream DiT layers.
Developers using Qwen-Image-2.1 can now automatically generate optimized prompts, reducing manual prompt engineering effort and potentially improving image generation quality. This may lower the barrier to using Qwen's image models for non-English speakers.
The model is available under the Qwen Research License Agreement, which may restrict commercial use. Businesses should verify the license terms before integrating it into production workflows.
The next signal to watch is whether Qwen releases similar prompt rewriting models for other modalities or integrates this capability directly into the Qwen-Image-2.1 pipeline.