Alibaba · Image model

Wan 2.7 AI Image Generator

Alibaba's Wan family is famous for video; Wan 2.7 brings the same visual intelligence to stills, with a twist: it thinks about composition before it renders. Two tiers — Standard for everyday work, Pro when the canvas goes up to 4K.

Output

Your generation will appear here.
Try one of these:

Session history

Nothing yet — your generations will appear here.

Misty river valley landscape at dawn — generated with Wan 2.7

01 · Overview

What is Wan 2.7?

Wan 2.7 is the still-image generation of Alibaba's Wan 2.7 family — the line best known for video, applied to photography-grade stills. Its text model runs a brief reasoning pass over your prompt (its "thinking mode") before rendering, which shows up as unusually coherent scenes: layered landscapes, believable crowds, props that belong together.

Every job comes in two tiers. Standard (5 credits) covers generation and edits at up to ~2K canvases; Pro (13 credits) switches to the pro endpoints, with canvases up to 4096×4096 total pixels for finals and print.

Standard 5 credits · Pro 13Thinking-mode compositionUp to 3 referencesPro canvases to 4K

Specs & capabilities

Inputs	Text to image · Image editing · Multi-reference (up to 3 in the studio)
Resolution	Standard tier up to ~2048×2048 total pixels; Pro tier up to 4096×4096
Aspect ratios	10 ratios in the studio, up to 21:9 ultrawide
Output formats	Set by the model (no format picker — Wan does not expose one)
Pricing shape	Standard endpoints flat $0.03, Pro endpoints flat $0.075 — priced here at 5 and 13 credits in both modes; reference images not billed separately.

Good to know: Wan's thinking pass makes turns slower than the speed models (typically 15–45 seconds). For rapid-fire drafting, Z-Image Turbo answers in seconds for 1 credit; Wan is where you send the prompts that need a considered composition.

02 · Features

Why creators pick Wan 2.7

Rainy futuristic city street at night with neon reflections — generated with Wan 2.7

Scenes that hold together

The reasoning pass pays off in busy frames: reflections agree with their light sources, crowds vary naturally, and foreground-to-background storytelling stays readable. Wan scenes feel directed rather than assembled.

Lantern festival poster rendered at the Pro tier — generated with Wan 2.7

A Pro tier when the canvas grows

This festival poster was rendered on the Pro tier: bigger canvas, more headroom for detail at display sizes. Draft the layout on Standard, then flip one toggle for the final — same page, same workflow.

A daytime street photo relit into a night scene — edited with Wan 2.7

Edits on both tiers

Day into night, season into season: this street was relit after dark from the daytime original. Edits take up to three references and are priced exactly like generation — 5 credits Standard, 13 Pro.

03 · Create

What you can create with Wan 2.7

Real outputs from this studio, with the exact prompts that made them. Load one into the tool and make it yours.

Documentary portrait of an elderly fisherman — generated with Wan 2.7

Documentary portraits

“A portrait of an elderly fisherman with a weathered face and kind eyes, harbor light, documentary photography”

Faces with life in them — Wan's photographic training shows in skin, light and expression.

Luxury watch product photograph — generated with Wan 2.7

Product drama

“A luxury watch product photograph on dark slate with dramatic side light and fine water droplets”

Moody, expensive-looking product stills — draft on Standard, print from Pro.

Scandinavian living room interior — generated with Wan 2.7

Interiors and spaces

“A bright Scandinavian living room interior with a green velvet sofa, oak floor and large windows, architectural photography”

Rooms that respect perspective and light — mood boards and listing visuals without a render farm.

04 · Guide

How to use Wan 2.7

Describe the scene

Give the thinking pass something to work with: subject, setting, light and mood. Or switch to Image to Image with up to three references.

Choose a tier

Standard (5 credits) for drafts and everyday finals; Pro (13 credits) when the image will be printed or displayed large — up to 4K total pixels.

Generate & download

Allow 15–45 seconds for the thinking pass. Credits are held at submit and charged only when the image succeeds.

Prompt guide

Write scenes, not keywords: the thinking pass rewards full descriptions — who, where, what light, what mood — with coherent compositions.
Direct the camera: “low angle”, “aerial view”, “documentary photography” — Wan's video heritage makes it fluent in camera language.
Draft Standard, finish Pro: lock the composition at 5 credits, then re-run the winner on Pro for the big canvas.
Relight with edits: “the same street at night, neon signs reflected in wet pavement” — time-of-day and season changes are Wan's home turf.
Be patient: the reasoning pass takes longer than turbo models; judge it on the frame, not the stopwatch.

05 · Pricing

Wan 2.7 pricing

Standard is 5 credits ($0.05) and Pro is 13 credits ($0.13) — the same in text and edit modes, references included. One-time credit packs, no subscription, credits never expire.

Mini

$4.90

490 credits · one-time

≈ 98 Standard images

Buy Mini

Starter

$9.90

990 credits · one-time

≈ 198 Standard images

Buy Starter

Pro

$19.90

1,990 credits · one-time

≈ 398 Standard images

Buy Pro

Studio

$39.90

3,990 credits · one-time

≈ 798 Standard images

Buy Studio

06 · FAQ

Wan 2.7 FAQ

How much does Wan 2.7 cost?

Standard generations and edits are 5 credits ($0.05 at the standard credit rate); the Pro tier is 13 credits ($0.13) in both modes. Reference images are never charged separately. New accounts start with 20 free credits.

What does the Pro tier add?

Pro runs on Wan's pro endpoints: canvases up to 4096×4096 total pixels (against ~2048×2048 on Standard) and more detail headroom for print and large displays. The workflow is identical — it is a quality toggle on the same page.

What is Wan's "thinking mode"?

Before rendering, the text model briefly reasons over your prompt — subject placement, light, scene logic — then generates. It is on by default upstream, which is why Wan's scenes hold together well and why turns take 15–45 seconds instead of a few.

How many reference images can an edit use?

Up to three in the studio, at no extra charge — edits cost the same as generation on each tier.

Why is there no format picker?

Wan's API does not expose an output-format option, so images arrive in the model's native format and are saved to your account history. If you need a guaranteed format, Z-Image Turbo offers JPEG, PNG and WebP.