Which image model does Text to PSD use?
The public image route is GPT Image 2 through KIE. It handles prompt-led generation and single-image editing.
Model guide
The image and video models behind Text to PSD
See which image and video models power Text to PSD, how provider fallback works, and which workflow fits your task.
- GPT Image 2 image route
- Editable PSD text layers
- Bounded provider fallback
- Credit estimate before the task
Server-renderedcontent, not a JS shell
Real credit costsshown before generation
9 language interfacefor the product shell
No fake previewssamples are labeled
The job
What this model guide is for
A useful creative tool is not just a model name in a dropdown. It is a routing decision: which model handles the image, which provider is healthy, what happens when the first provider fails, and how the result reaches an editable file. This page explains the model layer in plain language.
Images run through GPT Image 2 on KIE. Video runs through Seedance 2.0, with a provider fallback path that can use KIE and Volcengine at runtime, and adapters for fal, Replicate, APIXO, APIMart, and EvoLink. SeeAPI is used for content moderation, not image generation. WaveSpeed is present in the environment, but it is not a public generation route on this site today. We would rather say that clearly than pretend every provider is selectable.
Use this page when you are choosing between an image workflow and a video workflow, comparing credit cost, or checking what "multi-provider" actually means here.
01
Pick the right medium
Use a still image when the job is a poster, cover, menu, or banner. Use video when motion is the point, not a decoration.
02
Understand the credit cost
Image credits depend on resolution. Video credits depend on duration, resolution, fast mode, and optional inputs.
03
Know the failure behavior
Retryable provider failures can fall back. Policy rejections and invalid inputs stop, so you can fix the request instead of paying for a loop.
04
Plan the handoff
Images can move into the PSD studio. Video stays a media asset and can be paired with a separate cover or thumbnail.
What is actually wired
A model list you can verify
The public workspace only exposes routes that are connected and tested. This is the current split between image, video, moderation, and provider fallback.
| Layer | Route | Status | Best for |
|---|---|---|---|
| Image generation and editing | GPT Image 2 via KIE | Live | Posters, covers, menus, banners, product scenes |
| Video generation | Seedance 2.0 | Live | Short product, social, and property clips |
| Video provider fallback | KIE + Volcengine runtime; adapters for fal, Replicate, APIXO, APIMart, EvoLink | Retryable failures only | Keeping a request alive when an upstream is temporarily unavailable |
| Moderation | SeeAPI | Live | Prompt and media safety checks |
| WaveSpeed | Credentials configured, no public generation route | Not selectable | Not claimed on this page |
Image: GPT Image 2
Use GPT Image 2 for prompt-led image generation and single-image editing. It is the current public image route through KIE.
Best for posters, covers, menus, invitations, banners, and product scenes.
Video: Seedance 2.0
Use Seedance 2.0 for text-to-video, image-to-video, and reference-to-video. The runtime can fall back across providers when a request is eligible.
Best for short product clips, vertical social video, and listing teasers.
Moderation: SeeAPI
SeeAPI checks prompts and media before generation. It is a safety layer, not a creative model choice.
Keeps the public workspace aligned with content policy.
Fallback: KIE + Volcengine
The runtime starts with the configured video provider order and only falls back when the failure category is retryable. Content policy and input validation failures stop instead of retrying elsewhere.
Improves availability without hiding bad requests.
Live workspace
Try the image and video models behind text to psd
The model guide explains the image and video routes. Open the workspace that matches your deliverable.
Image and PSD workflow
Use GPT Image 2 for a still, then set editable text in the PSD studio. Best for posters, covers, menus, invitations, banners, and product scenes.
Open image workspaceVideo workflow
Use Seedance 2.0 for text-to-video, image-to-video, and reference-to-video. Best for short product clips and vertical social video.
Open video workspaceThe workflow
Four steps from brief to export
- 01
Start with the deliverable
Write down what you need to ship: a static poster, an editable layout, a short clip, or a set of ad variants.
- 02
Choose image or video
Use the image workspace for layouts and editable text. Use the video workspace when movement carries the message.
- 03
Check the cost
Open the pricing page or the generator estimate before you generate a batch. Retries and failed tasks should not be treated as free.
- 04
Export the right artifact
Download a PSD for design handoff, a PNG for review, or an MP4 for social and product pages.
Honest limits
Where this workflow stops
- Model availability can change. The public UI only exposes routes that are actually wired and tested.
- Provider fallback improves uptime but cannot fix a bad prompt, an unsupported input, or a content-policy rejection.
- A model name is not a quality guarantee. Review every output before publishing.
Before you publish
- Read the headline at phone size, not only on a desktop monitor.
- Check the product, face, or claim against the source.
- Confirm the credit cost and the plan before a batch.
- Reopen the exported PSD once before a handoff.
The image and video models behind Text to PSD FAQ
Direct answers about the workflow, the export, and what still needs a human review.
Which video model does Text to PSD use?
The public video workspace uses Seedance 2.0 for text-to-video, image-to-video, and reference-to-video.
Does the site use WaveSpeed?
WaveSpeed credentials exist in the environment, but it is not a public generation route in this build. The model guide lists only routes that are actually wired.
What is SeeAPI used for?
SeeAPI is the content moderation layer. It checks prompts and media; it does not generate images or video.
How does provider fallback work?
The video service tries the configured provider order when a failure is retryable, such as an upstream timeout. Policy and validation failures are not retried on another provider.
Can I choose a provider manually?
The public workspace chooses the route for you. Provider selection and fallback order are runtime configuration, not a user-facing model picker.
Why not list every model in the market?
Because a model list is only useful if the route works, the credits are correct, and the output can be reviewed. We keep the public list honest.
Where can I see the current price?
Use the pricing page for plans and credits. The generator also shows the estimated credit cost before you confirm a task.
Keep going
Related workflows and guides
Image workflow
AI poster maker with editable PSD text
Create poster concepts with GPT Image 2, then export a PSD with editable title and body text. Start from a prompt or photo.
Open workflow
Video workflow
AI product video generator from one photo
Turn a product image into a short video with Seedance 2.0, with clear controls for motion, duration, and aspect ratio.
Open workflow
Image workflow
AI YouTube thumbnail generator with editable headline text
Make high-contrast YouTube thumbnail concepts, then export a PSD with editable headline text for fast A/B testing.
Open workflow
Video workflow
AI social media video generator for vertical feeds
Create vertical video clips from a prompt or image with Seedance 2.0. Built for Reels, Shorts, and TikTok-first workflows.
Open workflow
9 min read
Image to PSD: What AI Can and Cannot Do in 2026
A realistic guide to turning an image into a PSD, including what automatic separation can do, what it cannot do, and a safer workflow for editable text.
Read guide
10 min read
Seedance 2.0 Image-to-Video: A Practical Guide
How to turn a still image into a short product or social video with Seedance 2.0, including motion prompts, duration, aspect ratio, and review checks.
Read guide
8 min read
Text to PSD vs Canva: Which Workflow Fits Your Team?
A practical comparison of a prompt-to-PSD workflow and Canva for posters, covers, menus, and campaign layouts, including where each one breaks down.
Read guideMake the first draft, keep the editable file.
Start with a prompt or a source image. Review the credit cost, generate a direction, then set the real copy in editable PSD layers.