Qwen Image 3.0 Pro text-to-image is the high-detail tier of Alibaba's Qwen line, spending more on multi-element typography and on skin, fabric and glass. It takes 5,000-character prompts at 1k or 2k in eight aspect ratios.
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.
Outputsakura-noir-temple-poster `prompt` describes the whole poster and quotes all three lines of type: the title SAKURA NOIR, the tagline EVERY PETAL HIDES A BLADE and the release line IN CINEMAS APRIL 4, 2027. All three come back letter for letter, each in the place and style the prompt asks for: large serif capitals across the top, a smaller italic line directly beneath, small spaced capitals along the bottom edge. Between them is the picture: a lantern-lit Kyoto temple courtyard with a pagoda, moss-covered stone lanterns and a storm of cherry blossom petals, and on the temple steps a woman in a crimson satin off-the-shoulder gown with a high slit, looking back over her shoulder at the camera with a half-smile. `aspect_ratio` `2:3` gives the poster shape, `resolution` is `1k`, the lower price tier, and `enable_prompt_expansion` is `false`, so the prompt reaches the model exactly as written.
- 1k
- 2:3
Outputarena-poster-three-lines Three separate short lines of type are named in the brief and set across the lower third, which is the case that separates a model that can letter from one that can only suggest lettering. `resolution` is `2k`, because at the lower tier a single short string survives but three stacked lines start to merge into each other. The type is kept out of the torchlit upper two-thirds on purpose - flat dark sand behind it is what keeps all three readable at once.
- 2k
- 2:3
Outputperfume-spread-tagline One line of type, set in negative space the brief reserves for it on the right, so the layout and the picture come back from the same call instead of from a render plus a later composition pass. `aspect_ratio` is `3:2` to give that column somewhere to live, and `resolution` is `2k` so a thin elegant face stays a letterform rather than turning into texture. The suspended mist is the second test: fine particulate held against backlight.
- 2k
- 3:2
Outputmagazine-cover The exact request behind the cover shown on this page: a 3:4 full-bleed portrait at 2k with masthead, issue line and cover lines all quoted in the prompt.
- 2k
- 3:4
Outputmotel-vacancy The second frame on this page: two separate strings — the neon script and the smaller amenity board beneath it — quoted in one prompt and both rendered legibly at 2k.
- 2k
- 3:4
Outputmorning-note-pov The other two frames on this page put their words on printed matter; here the words are handwritten in blue ink on a note propped against a cup, and the camera is the person who has just woken up. Dashes, a full stop and a signed initial all survive, and they sit behind the plane the lens is focused on.
- 2k
- 3:2
Outputprivate-eye-door-glass - 2k
- 4:3
Outputsalt-crystal-fragrance - 2k
- 3:4
This example starts at $0.04 per image. The Playground updates the estimate when you change the settings.
