Skip to content

Picsart Opens 250+ AI Models To Developers Through One API

Picsart, which says more than 100 million people use its editing app each month, now sells developers pay-per-use access to over 250 image, video, audio and text models through a single connection.

Picsart Opens 250+ AI Models To Developers Through One API
Image courtesy: Picsart

Developers and businesses can now use the AI behind Picsart’s photo and video editing app in their own products. Picsart has launched Picsart Platform, which offers more than 250 AI models from over 30 providers, plus its own editing models. They all work through one application programming interface (API), a link that lets one piece of software talk to another.

Picsart was founded in 2011 in Yerevan, Armenia, and is now headquartered in Miami, US. It says more than 100 million people use its app each month and that it has been downloaded more than 2.5 billion times. SoftBank Vision Fund 2 led a $130M round in Picsart in 2021 that valued the company at more than $1B, and Sequoia Capital is also a backer.

One Connection For Hundreds Of Models

New AI models for video and images now arrive often, and teams using several of them have to manage separate connections, access keys and bills. Picsart says its platform replaces that with one integration and gives access to the latest models as soon as they launch.

The models come from providers including Google (Veo), OpenAI, ByteDance, Kling, Runway, ElevenLabs and Alibaba, according to Picsart’s platform page. The company says developers can switch between models without rewriting their code.

Picsart says its own editing models, such as Enhance and background removal, cannot be used anywhere else. Developers can use them to edit images that other models produce.

“Generative media is moving faster than any team can integrate it, our job is to keep up so our customers don’t have to,” said Hovhannes Avoyan, Picsart’s founder and chief executive.

Built For Developers And AI Assistants

Developers can reach the platform in several ways. These include a web API, ready-made code libraries known as software development kits (SDKs) for Python and TypeScript, a command-line interface (CLI) for typing commands directly and a playground for testing models.

Picsart also offers Picsart Media Studio, a server based on the Model Context Protocol (MCP), an open standard that lets AI assistants use outside tools. According to the platform page, the server connects assistants such as Claude Code and Cursor to Picsart’s model catalogue.

Self-serve users pay for each generation, with no seat fees or subscription minimum, according to the platform page. It lists Picsart Image Edit from $0.01 per generation and Kling 4.0 from $0.157 per second of video. As a launch offer, new users had unlimited access to Alibaba’s Wan 3.0 video model from 20 to 22 November.

A Crowded Field For Model Access

Picsart is not the only company offering many models in one place. San Francisco-based each::labs, which raised $7M in seed funding this month, offers more than 600 AI models for images, video and audio through a single API.

Avoyan says the aim is to give companies of every size the same tools. “A two-person startup and a global brand should have access to the same creative engine,” he said.

With the launch, Picsart is selling the technology behind its consumer app to other builders, in a market where leading models change often.

Add Morning Tick on Google