Qwen Image 2 is Alibaba's text-to-image model, reading prompts of up to 800 characters natively in Chinese or English. It returns one image per request in 1:1, 4:3, 3:4, 16:9 or 9:16, with an optional seed.
Workbench
These examples were written for this endpoint. Replace the prompt and sample media URLs with your own accessible inputs, then check the current quote and model availability before submitting.
Outputcyber-alley-bilingual The brief has to fit inside the 800-character `prompt` ceiling and still carry two scripts: the neon signage is written out in both Chinese and Latin characters, so the render is judged on lettering as well as on the wet-shirt highlight it throws. `aspect_ratio` is `3:4` because the composition stacks - signage above, the wiped patch of condensation at eye level, the figure behind it. Note how little of the wording goes on the subject and how much of it on the two colour temperatures cutting the silhouette.
- 3:4
Outputmidnight-fox-bar The exact request behind the still on this page: one square brief covering wardrobe, a tail accessory, the bar set and the lantern light.
- 1:1
Outputclub-dj-illustration The second frame on this page, and a demonstration of how much of the 800-character budget a full illustration brief eats: style, character, wardrobe, framing, lighting and an explicit no-text instruction.
- 1:1
Outputdo-not-knock-corridor The one still on this page with words in it. The brief spells out a brass door plate reading 214 and a card hanger reading DO NOT KNOCK, and both landed legibly at 3:4 — short strings sometimes do. In-image text is not this family’s strength, though: for signage, mastheads and packaging copy, reach for the Qwen Image 3.0 line.
- 3:4
Outputlantern-festival-yukata - 3:4
Outputbrutalist-library-rain - 3:4
Outputsupper-club-singer - 3:4
This example starts at $0.035 per image. The Playground updates the estimate when you change the settings.