Image generation model by Alibaba. Up to 4K output, multi-image editing with 9 references, thinking mode, HEX color palette control, and text rendering in 12 languages
Photorealistic portrait of a young woman with long red hair flowing in the wind, against a sunset over the ocean
Standard for everyday generation and editing, Pro for maximum quality text-to-image
Full-featured generation and editing pipeline — text-to-image, multi-image input, thinking mode, and color palette control. Output up to 2K.
Everything in Standard plus 4K output for text-to-image generation. Same multi-image editing, thinking mode, and all creative controls.
Detailed comparison between Standard and Pro versions
| Parameter | Standard | Pro |
|---|---|---|
| Input images | Up to 9 | Up to 9 |
| Output size | 1K, 2K | 1K, 2K, 4K (T2I) |
| Outputs per request | 1–4 (up to 12 gallery) | 1–4 (up to 12 gallery) |
| Thinking mode | Yes (T2I) | Yes (T2I) |
| Interactive editing boxes | Yes | Yes |
| Color palette | 3–10 HEX colors | 3–10 HEX colors |
| Text rendering | 12 languages | 12 languages |
| Aspect ratios | 8 options | 8 options |
Unified generation and editing pipeline with thinking mode and precise creative control
Text-to-image and image editing in a single flow — generate from scratch or transform existing images seamlessly. No switching between modes or APIs.
Fine control over facial bone structure, eyes, contours, makeup, hairstyle, and accessories for detailed character creation. Ideal for avatar systems and character-driven content.
Specify 3-10 HEX colors with weight ratios for exact color scheme matching. Extract palettes from reference images — perfect for brand guidelines and design systems.
Accurate long text in 12 languages including tables, formulas, and infographics. Wan 2.7 Image handles complex typography far beyond typical AI image models.
Upload up to 9 input images with region-based box selection for precise local edits. Specify up to 2 bounding boxes per image — targeted changes without affecting the rest.
Generate up to 12 stylistically consistent images per request in gallery mode. Pro version supports 4K output for maximum print and display quality.
From prompt to result in 5 steps
Select Wan 2.7 Image (Standard or Pro) from the model catalog on the image generation page.
Describe the image in detail — subject, environment, style, lighting. Upload up to 9 reference images if needed.
Pick aspect ratio, resolution (1K/2K/4K), enable thinking mode, set color palette, or define editing regions.
Hit create — Wan 2.7 Image processes your request in about 45 seconds with precise text, colors, and realistic details.
Download the result in original quality. Use it as a base for further edits or as a reference for consistent image series.
How Wan 2.7 Image benchmarks against leading image generation models
Compared against GPT Image 1.5, Seedream 4.5, Kling Image 3.0, Seedream 5.0 Lite, and Nano Banana Pro
Identity preservation, advanced editing, multi-image handling, local editing, text editing, and style control
Built for creative professionals and teams
Generate multiple visual variations for A/B testing — palette-matched to brand guidelines with up to 12 consistent images per batch.
Detailed character creation with bone structure, makeup, and accessory control — ideal for gaming, social platforms, and personalized content.
4K output for print-ready marketing materials, packaging mockups, and high-resolution displays — exact HEX color matching included.
Gallery mode generates up to 12 stylistically consistent images per request — perfect for storyboards, lookbooks, and product catalogs.
The best way to use Wan 2.7 Image
Pay per generation with clear credit costs — no hidden fees, no API key management, no usage surprises. Standard from 2 credits, Pro from 4.
Full parameter control through an intuitive UI — aspect ratio, resolution, thinking mode, color palette, reference images, and editing regions.
Automatic failover across providers, queue management, and CDN delivery — your generations complete reliably without API downtime.
Answers to popular questions about Wan 2.7 Image
Wan 2.7 Image is an image generation model by Alibaba. It supports text-to-image generation up to 4K and image editing with up to 9 reference inputs, thinking mode, color palette control, and text rendering in 12 languages.
Standard supports 1K and 2K output for both T2I and editing. Pro adds 4K output but only for text-to-image generation. Both support multi-image input, thinking mode, and all aspect ratios.
Upload up to 9 images and reference them in your prompt using @image1 through @image9. You can also specify up to 2 bounding boxes per image for region-based editing — targeted changes without affecting the rest.
Thinking mode enables deeper reasoning about scene composition, spatial relationships, and complex prompts. It's available for T2I without gallery mode — ideal for intricate multi-element scenes.
Specify 3-10 HEX colors with percentage ratios (summing to 100%). The model matches these exact colors in the output — perfect for brand guidelines and design consistency.
Wan 2.7 Image renders accurate text in 12 languages including English, Chinese, Japanese, Korean, and European languages. It handles long text, tables, formulas, and infographics.
Typical generation takes about 45 seconds. Results arrive automatically — you can continue working during generation.
