Intro: One API, Many Models

What is the AI Providers Hub?

The AI Providers Hub is a unified layer in Picsart Creative APIs that lets you select which model runs a given service (e.g., Text‑to‑Image) while keeping the same endpoint, auth, and request pattern. Prefer a different vendor for a particular job? Change the model identifier, and your integration stays the same.

Highlights

Single integration, multiple models. Stay on Picsart; no new SDKs, keys, or platforms.
Vendor‑neutral. Use Picsart’s native models or third‑party options (e.g., Flux Kontext Max) for the same workflow.
Faster evaluation. A/B test quality, speed, or cost by toggling models, not rewriting code.

How model selection works

All Hub‑enabled services share a unified request shape. You choose the model via a standard AIR URN and (optionally) pass provider‑specific options in a scoped object. Everything else in your request is unchanged.

model=urn:air:picsart:model:picsart:sod@9

{
  "model": "urn:air:sdxl:model:fluxai:flux_kontext_max@1"
}

urn:air:sdxl:model:fluxai:flux_kontext_max@1 urn:air:picsart:model:picsart:fluxt2i@1

📔
Notes
Picsart uses AIR (AI Resource) identifiers defined by CivitAI to reference models.
AIR URNs are stable, explicit, and versionable to support reliable deployments.

Model identifiers (AIR URNs by CivitAI)

Picsart adopts the AIR naming scheme to uniquely and consistently identify models across providers.

Examples

urn:air:picsart:model:picsart:sod@9
urn:air:picsart:model:picsart:sod@10

Why AIR?

Clarity. URNs are self‑describing and vendor‑agnostic.
Version control. Pin a model version with @{version} for reproducible outcomes.
Interoperability. A common format simplifies cross‑provider tooling and governance.

You can switch versions or providers by changing only the URN, keeping prompts, dimensions, and job handling the same.

Switching models: a one‑line change

Pinned version to pinned version:

- "model": "urn:air:picsart:model:picsart:sod@9"
+ "model": "urn:air:picsart:model:picsart:sod@10"

Across providers, you use the same pattern. Replace the URN with the provider/model/version you want to evaluate.

Normalized parameters vs. provider‑specific features

The Hub normalizes common parameters so you don't have to relearn basics for every provider. When a model exposes unique capabilities, you can opt into them explicitly.

Normalized across providers (examples)

Input/Output basics: prompt, negativePrompt (when supported), width, height, seed
Content type: input image or mask fields for image‑to‑image workflows
Job control: idempotency keys, webhooks/callbacks, status polling

Provider‑specific (examples)

Advanced controls (e.g., prompt enhancement, motion/morph controls)
Extended ranges (e.g., max resolution, clip duration)
Experimental features or schedulers

Capabilities & constraints to expect

While the Hub smooths differences, each provider still has real limits. Always check the provider page for details on:

Supported modes. Not every provider supports every workflow (Text‑to‑Image, Image‑to‑Image, Video, etc.).
Parameter ranges. Aspect ratios, max resolution, steps/quality settings, clip duration can vary.
Input requirements. Accepted file types, size limits, recommended color spaces may differ.
Feature availability. Some models add audio, style presets, or safety filters that others don’t.

Versioning & stability

Models evolve. To control upgrades:

Pin specific versions with AIR URNs (e.g., …@9, …@10) in production.
Use preview or rolling channels only if explicitly documented for a provider.
Track each provider’s changelog for deprecations and breaking changes.

Example: Text‑to‑Image on multiple providers

Picsart Text‑to‑Image: Balanced quality, speed, and style presets tuned for common design tasks.
Third‑party options (e.g., Flux Kontext Max): Alternative looks or tuning preferences for certain creative briefs.

Pick what fits the job; keep the same API surface and deployment patterns.

Request a new model

Don’t see the model you want? We're expanding the Hub continuously. Request an integration and share the provider, AIR URN (if known), and your use case. We’ll review fit, licensing, and timelines and follow up.

👉 Request a model via our contact form

(If your team must use a specific vendor contract or BYO key, mention it in the request. We’ll let you know what’s possible.)