Choose Wan2.7 Image for routine concept and edit work
Wan2.7 Image is the straightforward starting point when 1K or 2K output is enough. That covers many social concepts, moodboards, interface illustrations, storyboard frames and reference-led edits. Beginning with the base route also lets you validate prompt structure, composition and style before spending additional Credits on resolution you may not need.
Resolution is not the same as quality, accuracy or suitability. A larger file will not repair unclear instructions, conflicting references or a composition that fails the brief. Evaluate the subject, typography risk, hands, brand details and edit fidelity at the intended display size rather than assuming Pro wins every task.
- Use 1K for rapid exploration when the final channel allows it.
- Use 2K when review detail or modest cropping requires more pixels.
- Keep references visually consistent and give each one a clear purpose.
- Retain the prompt and settings beside every candidate for repeatability.
Choose Image Pro when eligible 4K output changes the deliverable
Wan2.7 Image Pro adds 4K output for eligible text-to-image generation. That can matter when a large digital canvas, substantial crop or high-resolution master is part of the brief. The official API warns that 4K is not universal: reference-based and editing scenarios remain limited to the supported lower resolution options.
Do not select Pro solely because the label sounds better. First ask whether the finished asset needs native 4K, whether the scenario qualifies and whether a 2K result could be reviewed before generating the final master. Because Token Plan Credits are dynamic, compare actual consumption in the console rather than inventing a price ratio between the models.
Sequential generation can create a coherent set
The Wan image guide documents sequential generation for producing between one and twelve related images. This is useful for storyboards, campaign variants and visual sequences where continuity matters more than isolated prompts. It can reduce manual prompt drift, but every additional output still needs review and can affect overall Credits usage.
Some controls are mutually exclusive. When sequential generation is enabled, the documentation says thinking mode and the colour-palette control are unavailable. Decide which form of control is more important before sending the task. For brand work, a short controlled sequence may be more useful than requesting the maximum number of frames.
- Use enable_sequential when the images should read as one set.
- Request only the number of candidates needed for the decision.
- Use a palette of three to ten colours when palette control is available.
- Separate concept approval from final-resolution generation.
Token Plan integration is separate from ordinary chat
Alibaba's Individual quick start says image and video generation models use a separate API and should be connected through a compatible AI tool's Skill or extension mechanism. The dedicated Token Plan key and Base URL must remain paired. A chat client that can call Qwen text models does not automatically expose Wan image generation without the additional integration.
Generation is asynchronous: create the task, preserve its ID, poll the status responsibly and retrieve the output after success. Within Token Plan Individual, keep that flow interactive and personal. The plan terms prohibit using the key in a public app backend, automated script or non-interactive batch.
Who this is for—and who it is not for
Choose wan2.7-image when 1K or 2K is sufficient and you want to iterate on concepts or edits. Choose wan2.7-image-pro when an eligible text-to-image job genuinely needs a 4K master. Both suit an individual creator working interactively through a supported agent or creative-tool extension.
Neither model turns the Individual subscription into a stock-image factory or customer-facing API. Do not promise a fixed number of images per plan, and do not infer that a versioned Wan model is included because its name is similar. Check exact IDs, current terms and actual Credits consumption before committing.