In short: Most AI image tools hand you one flat picture. Prisma V1.0 hands you a design you can keep working on: live text, native shapes and separate image layers, returned as editable Fabric.js-format JSON through a self-serve API. There’s no subscription: one-time refills start at $10 and last 12 months.
The problem with flat AI pixels
Image generators are very good at making pictures. But a marketing design isn’t just a picture. It’s a headline legal will want to change, a price that changes next week, a logo that must sit exactly in its corner, and the same layout in ten languages.
When an AI model returns a PNG or JPG, all of that is fused into pixels:
- Text is part of the image. A typo, a new price or a translation means generating again and hoping the layout survives.
- Nothing can be moved. The headline, the product photo and the background are one layer.
- Brand control is hard. Your exact font, colour or logo can’t go in without rebuilding the design by hand.
So teams use AI for inspiration, then rebuild the layout in an editor. Where layered AI output does exist, it usually stays inside a closed app or comes with an enterprise contract, not as an API developers can build on.
Introducing Prisma V1.0
Prisma V1.0 is a managed, asynchronous design API from AI Smart Content, released in early access on 9 October 2026. You send a prompt. Prisma creates a complete design, then rebuilds it as editable layers that your product, your team or your customers can change.
| Flat AI image | Prisma V1.0 | |
|---|---|---|
| Output | One PNG or JPG | Editable layered JSON plus private image layers |
| Text | Baked into pixels | Live, editable text |
| Shapes | Pixels | Native, editable shapes |
| Photos you supply | Redrawn or lost | Placed untouched as separate images |
| Fixing a typo | Generate again | Edit the text |
| Access | An app or an enterprise deal | Self-serve API keys |
How Prisma works, in three steps
Prisma combines PIC Studio’s tested goal-guided generation and editable reconstruction into one managed service. Every job follows three steps:
- Generate the goal. Prisma first creates a complete visual goal for your prompt: the whole design as it should look.
- Rebuild editable layers. The design is rebuilt against that fixed goal. Text and native shapes are separated from raster imagery, so the parts you’ll want to change stay editable.
- Review and retrieve. The rebuilt design is compared with its unchanging goal, and you download the result. Every output stays reviewable rather than automatically approved for production.
How Prisma turns a flat goal into clean layers is our own recipe, so we’ll leave that part behind the scenes. What matters to you is what comes out.
What you get back
- Editable JSON in the legacy-fabric-v1 format. It opens in Fabric.js and can be parsed on your backend; the docs include PHP and Node.js examples.
- Live text you can retype, translate or restyle.
- Native shapes that stay sharp and editable.
- Separate image layers as private PNG files, plus a JPEG preview.
- Your own photos, untouched. Send up to five material images and Prisma places each one as its own editable image.
Fonts may be substituted and fine details may need a touch-up. Treat each result as a strong first draft to review, not a finished asset.
Want to see a result before writing code? After a Playground generation, Prisma opens that template in a private editor sandbox. You can edit it, ask AI Smart Magic to apply changes live in Fabric.js, and download the JSON. The sandbox needs a sign-in and a paid refill.
Built for developers
Prisma is a plain HTTPS API:
- Scoped keys. Create keys in the developer portal and send them as Bearer tokens. Scopes are generations:write, jobs:read and assets:read. Keys expire, can be revoked and are shown only once.
- Asynchronous jobs. POST /api/v1/generations returns 202 Accepted with a job ID. Poll GET /api/v1/jobs/{job_id} every few seconds, then fetch /result for the editable JSON.
- Safe retries. Every request carries an Idempotency-Key, so repeating a request never charges twice.
- Five sizes. 1:1, 4:5, 9:16, 16:9 and 1.91:1.
- Creative controls. Prompts of 20 to 1,600 characters; an optional reference image, used for its mood or its style; native circle and oval frames; device and merchandise mockups such as phone, laptop, T-shirt, mug and poster.
- Clear limits. 60 API requests and 5 new generations per minute per key; 2 simultaneous generations and 10 active keys per workspace.
A minimal request:
curl --fail-with-body 'https://picstudio.aismartcontent.co/api/v1/generations' \
-H "Authorization: Bearer $PRISMA_API_KEY" \
-H 'Content-Type: application/json' \
-H 'Idempotency-Key: campaign-2026-001' \
-d '{"model":"prisma-v1.0","prompt":"An original matcha campaign with a large heading and separate product photography.","name":"Matcha launch","aspect_ratio":"4:5"}'The documentation covers every field, job state and error code, with cURL, Python and JavaScript examples.
Pricing: no subscription, just refills
Prisma runs on Prisma Credits, and each generation costs 500. Refills are one-time purchases with no monthly plan, and every refill includes private cloud storage for 12 months.
| Refill (USD) | Prisma Credits | Generations | Private cloud storage |
|---|---|---|---|
| $10 | 2,500 | 5 | 1 GiB |
| $20 | 6,000 | 12 | 2 GiB |
| $35 | 11,000 | 22 | 5 GiB |
| $50 | 16,000 | 32 | 10 GiB |
| $100 | 32,500 | 65 | 25 GiB |
| $200 | 65,000 | 130 | 50 GiB |
| $500 | 165,000 | 330 | 150 GiB |
| $1,000 | 332,500 | 665 | 300 GiB |
| $2,000 | 667,500 | 1,335 | 500 GiB |
- Credits are valid for 12 months from the refill.
- Any whole-dollar amount from $10 to $5,000 also works, at the rate of the biggest refill it covers.
- Credits are reserved when a job is queued and charged only after a saved result. If a generation fails, the reservation is released.
- API keys cost nothing extra. Prices are in USD before any applicable tax.
See the current table on the pricing page.
Who Prisma is for
- SaaS and app builders who want a “generate a design” button whose results their users can edit.
- Agencies and studios producing many on-brand variations for campaigns and clients.
- E-commerce and marketplace teams that need product visuals with real, editable text and prices.
- Social and content teams working across many sizes and languages.
What V1.0 doesn’t include yet
Early access means being clear about limits. Prisma V1.0 offers no downloadable model weights, SDK package, webhook delivery or SLA. Review each output before use, and use source images and fonts you have the rights to.
Get started
- Sign in at the Prisma developer portal and choose a workspace you own.
- Create an API key and keep its secret on your server.
- Refill Prisma Credits.
- Submit a design, poll the job and download your editable result.
FAQ
What is Prisma V1.0?
A managed, asynchronous design API from AI Smart Content. It generates a complete visual goal from your prompt, then rebuilds it as editable text, shapes and image layers.
What format is the output?
Editable JSON in the legacy-fabric-v1 format, which opens in Fabric.js, plus private PNG image layers and a JPEG preview.
Can I edit the text after generation?
Yes. Text and native shapes come back as editable layers, not pixels. Fonts may be substituted, so check each result.
Is there a monthly subscription?
No. Prisma uses one-time refills, from $10 for 2,500 credits (5 generations). Credits last 12 months.
How much does one design cost?
500 Prisma Credits per generation. If a generation fails, the reserved credits are released.
Which sizes are supported?
1:1, 4:5, 9:16, 16:9 and 1.91:1.
Is there an SDK or webhooks?
Not in V1.0. You call the HTTPS API directly and poll each job; the docs include cURL, Python and JavaScript examples.