Bria AI API
Access Bria AI instantly with Puter.js, and add AI to any app in a few lines of code without backend or API keys.
// npm install @heyputer/puter.js
import { puter } from '@heyputer/puter.js';
puter.ai.txt2img("A beautiful sunset", {
model: "bria/fibo"
}).then(imageElement => {
document.body.appendChild(imageElement);
});
<html>
<body>
<script src="https://js.puter.com/v2/"></script>
<script>
puter.ai.txt2img("A beautiful sunset", {
model: "bria/fibo"
}).then(imageElement => {
document.body.appendChild(imageElement);
});
</script>
</body>
</html>
List of Bria AI Models
FIBO
bria/fibo
FIBO is a text-to-image model from Bria AI, launched in November 2025. Rather than encoding a prompt directly, an accompanying language model expands a short prompt into a structured JSON specification covering visual attributes such as lighting, camera angle, composition, and color, then generates the image from that specification and returns the JSON alongside it. FIBO has 8 billion parameters and is trained on structured JSON captions running to 1,000+ words, using licensed images from more than 20 content partners, including Getty Images and Envato. FIBO also supports a refine mode that changes only the attributes named in a follow-up instruction while holding the rest of the specification fixed, and an inspire mode that derives a JSON specification from a reference image. Bria reports it leads open alternatives on image-text alignment and aesthetics in PRISM-style evaluations, suited to teams that need reproducible output such as brand or product imagery.
ImageImage 3.2
bria/image-3.2
Image 3.2 is Bria AI's text-to-image model for commercial use, an update to Bria 3.1 in the company's Bria 3 series. It runs on a 4 billion parameter rectified-flow transformer with a T5 text encoder, about a third the size of the 12 billion parameter open-weight models it's compared against, while Bria reports aesthetics and prompt adherence on par with those larger models. The model is trained entirely on licensed image data, avoiding copyrighted characters, logos, trademarks, and public figures in its training set. Text rendering was a focus for this release. Bria's own benchmark shows OCR accuracy for short in-image text (one to six words) rising from 5% in version 3.1 to 70% in version 3.2. The smaller parameter count also brings faster fine-tuning, about 2x quicker on L40S and A100 GPUs, useful for teams adapting the model to a specific brand or product style.
Frequently Asked Questions
The Bria AI API gives you access to models for AI image generation. Through Puter.js, you can start using Bria AI models instantly with zero setup or configuration.
Puter.js supports a variety of Bria AI models, including FIBO and Image 3.2. Find all AI models supported by Puter.js in the AI model list.
With the User-Pays model, users cover their own AI costs through their Puter account. This means you can build apps without worrying about infrastructure expenses.
Puter.js is a JavaScript library that provides access to AI, storage, and other cloud services directly from a single API. It handles authentication, infrastructure, and scaling so you can focus on building your app.
Yes — the Bria AI API through Puter.js works with any JavaScript framework, Node.js, or plain HTML. Just include the library and start building. See the documentation for more details.