Qwen Image 2.1 Pro vs Turbo vs Qwen Image 3, Nano Banana 2.1 and GPT Image 2.5

Qwen Image 2.1 is Alibaba's compact image generation and editing series, built on a 7B visual generation model. It takes up to 10 reference images for editing.

• Qwen Image 2.1 Pro is the higher-quality tier. Alibaba positions it for a balance of speed, quality and cost, with Qwen Image 3.0 Pro as the flagship. Pro can also decide from the prompt whether to output an image with a transparent background.
• Qwen Image 2.1 Turbo is the accelerated open-weights checkpoint. It generates in 8 denoising steps at up to about 2.7K on the long side.

Pricing (per 1M image-output tokens; text input is free on all Qwen models):

• Qwen Image 2.1 Turbo $4.41, about 2.5× cheaper than 2.1 Pro
• Qwen Image 2.1 Pro $11.02, which is more per token than Qwen Image 3 Pro ($9.58) and Qwen Image 3 ($7.19)
• Qwen Image 3 and 3 Pro add a flat $0.003 per image
• For comparison, Nano Banana 2.1 and GPT Image 2.5 Flare both cost $30, roughly 3× Qwen 2.1 Pro and 7× Turbo. They also bill input ($1.50 and $8.00 per 1M).

Choosing between them: Turbo is the cheap, high-throughput option you can also run locally. Choose 2.1 Pro for multi-reference editing and transparent output, not for price. The Western models cost far more per output token, so test them on your own prompts before paying for them.

Caveats: the 2.1 series does not accept a negative prompt. Turbo's weights are under the Qwen Research License Agreement, so check its terms before commercial use. Tokens per image scale with resolution.

Qwen-Image-2.1 (Qwen blog) · Qwen-Image-2.1-Turbo weights