Stop Searching,
Start Building.

The Developer-First Hub for Open-Source AI Workflows

Text Link
Highlighted Blueprint
Made by 
Mozilla.ai
WASM Agents

This Blueprint demonstrates how to run AI agents in your browser using WebAssembly (WASM) through Pyodide and the OpenAI Agents Python SDK.

Build an Agent
Browser
Mozilla.ai
Preview this Blueprint in action
00:00
00:00
Blueprint Related Content

Want to build your own Blueprints?

See our guidelines for building a top-notch Blueprint.

Must-haves

Open-source models and tools usage

README, pyproject.toml, and organized folder structure

Demo app (Streamlit or Gradio) or jupyter notebook

Config file for easy customization

CLI support

Nice-to-haves

CPU compatibility for most local setups

Google Colab notebook option

PyPI package availability

Dockerfile for the demo app

Diagram of the Blueprint in the README

Setup and guidance docs using mkdocs

Highlighted Building Blocks

Explore the open-source resources behind our Blueprints.

Tools
Any-Agent

Any-Agent is a Python library designed to provide a single interface to access many different agent frameworks.

Whisper BiDec

Whisper BiDec enables to adjust transcriptions and recognize unusual names or phrases with smaller Whisper models.

Ultralytics

Ultralytics provides cutting-edge computer vision models, including YOLO11, enabling developers to integrate real-time object detection, segmentation, and classification into AI applications with minimal effort.

Datasets
Common Voice
Common Voice

A multilingual, crowdsourced collection of voice recordings from Mozilla.

Alpaca-gpt4

This dataset contains English Instruction-Following generated by GPT-4 using Alpaca prompts for fine-tuning LLMs.

Models
OuteTTS-0.1-350M

OuteTTS-0.1-350M is a novel text-to-speech synthesis model.

Qwen2.5-3B-Instruct-GGUF

Qwen2.5-3B-Instruct-GGUF is an instruction-tuned model that generates long-form content, and is optimized for efficient deployment via the GGUF format.

Kokoro-82M

Kokoro is an open-weight TTS model with 82 million parameters.