What is the Picsart API and how to use it

The Picsart API brings 180 AI models from 32 providers into one place. Image, video, audio, and text models all work the same way, so trying a different model means naming a different one and changing nothing else. The whole catalog is reachable from a project, from a terminal, or from an AI assistant.

What the Picsart API platform covers

One catalog, one account, and one way of paying: per generation, with no servers to run.

• 180 models
• 32 providers
• 4 modalities
• Pay per generation

All 180 sit in the Picsart model catalog. Pick any model there to see its settings and ready-to-copy examples.

Who the Picsart API is for and what it does

The platform suits teams that need generation inside something they are already building, rather than a place to go and make one asset by hand. That means product and app teams, teams working in any programming language, content and automation teams working from a terminal, and anyone building on AI assistants. None of them has servers to run.

These are the jobs the catalog covers.

• Product and ecommerce imagery. Remove a background or swap it for a new scene described in words, change the camera angle on a subject with the lighting corrected automatically, and sharpen the result.
• Social and ad creative. Make images from a written description, apply curated preset effects, and animate a still photo into video.
• Video production. Make video from text, images, or existing footage, join clips into one continuous cut, restyle what already exists, and reframe it to a new aspect ratio.
• Voice and music. Create speech in many languages, copy a voice from a sample recording, dub audio and video into other languages, and generate music with vocals or instrumental, plus custom sound effects.
• Talking avatars. Turn a portrait and a voice track into a video that lip-syncs.
• Text and understanding. Write text, and read an image or a video back as text.

Nothing on that list needs a separate account with a separate vendor, because every one of those jobs is a model in the same catalog.

Four ways to reach the catalog

Every model works the same way. Pick whichever route matches how the team already builds.

• A ready-made toolkit. One instruction generates anything in the catalog, with every model and setting spelled out as it gets typed.
• A direct web request. The same jobs, from any programming language.
• A terminal tool. Generate, script, and automate from the command line, with cost quotes.
• A connection for AI assistants. Claude Code, Cursor, and other compatible assistants can generate on their own.

Three things stay the same whichever route gets used:

• The same settings work across all three
• A cost quote arrives before a credit is spent
• Quick models answer straight away, and longer jobs report back once they finish

Images, video, audio, and text in one place

All four come from a single account.

• Image, 66 models. Make and edit images.
• Video, 83 models. Make video from text, images, or video.
• Audio, 23 models. Speech, music, and sound effects.
• Text, 8 models. Writing and understanding.

Every model says up front what it takes in and what it hands back, so the shape of a job is clear before reading any settings. Between them the 180 models turn text into images, images into new images, text into video, images into video, video into new video, audio into video, text into speech, one voice into another, text into music and sound effects, and images or video into text.

Picsart’s own models sit in that catalog too, covering background removal and replacement, sharpening and face enhancement, general image editing, camera angle changes with automatic relighting, virtual makeup, fast image generation, and curated preset effects for photo and video.

Filter by type in the AI model catalog to see which models handle which.

Start building in minutes

Step 1. Get a key

Create an API key and save it with the project. The toolkit reads it from there, so it never needs typing again into individual requests.

Step 2. Choose how to connect

The ready-made toolkit is the quickest route for a TypeScript project, because it spells out every model and setting as they get typed. Teams working in another language can skip it and send the same jobs as ordinary web requests.

Step 3. Make the first image

Point the request at a model, describe the image in a prompt, and the platform hands back a link to the finished file. That is the whole first run.

Step 4. Try a different model

Name another model from the catalog and send the same request again. Everything else stays put, because every model in the catalog works the same way.

The quickstart walks through that path end to end, and the API reference covers every option in detail.

Start building with the Picsart API

Grab a key, run one generation, and swap in a second model to see how little changes. Get started on the Picsart API platform.