Mozilla.ai ORG

Distribute and run LLMs with a single file.

Communicate with an LLM provider using a single interface

An open standard for shared agent learning. Agents persist, share, and query collective knowledge so they stop rediscovering the same failures independently.

A single interface to use and evaluate different agent frameworks

Source code for Mozilla.ai's Lumigator platform

Open-source, OpenAI-compatible LLM gateway you run yourself. One endpoint for 40+ providers, with virtual keys, budgets, and usage tracking.

Testing WASM-powered AI agents

Building blocks for agents in C++

Blueprint by Mozilla.ai for generating podcasts from documents using local AI

Declaratively define and run required tools across environments, from local development to containerized cloud deployments.

any-llm go

Distribute and run transformer encoders with a single file.

Giving researchers and practitioners an easy way to switch between guardrails.

Generate agents by simply describing your task in natural language

The AI Assistant that actually does things for the trades

Blueprint by Mozilla.ai for mapping features in OpenStreetMap with Computer Vision

Blueprint to Build Your Own Timeline Algorithm

Blueprint by Mozilla.ai for finetuning a Speech-To-Text model in your own language

Your buddy in the (L)LM space.

Convert unstructured documentsto markdown using the Docling.

Blueprint for federated finetuning, enabling multiple data owners to collaboratively fine-tune models without sharing raw data. Developed in collaboration with Flower.

Blueprint by Mozilla.ai for answering questions about structured documents

Template that can be used to start your own Blueprint.

No description.

Python SDK for mcpd

Blueprint by Mozilla.ai on how to transcribe audio files

Minimalist framework for running and monitoring an agent

mcpd proxy MCP server for IDE integrations

No description.

A simple command-line tool to calculate importance of tokens in prompts sent to an LLM.

Hex Settlement Game: TUI, written in Rust

Visual demo of DSPy's prompt optimization on Gradio

Aligning / Biasing Speech-to-Text models to a predefined set of words or phrases.

Blueprint for training and deploying a machine learning model that effectively detects synthetic and modified audio content.

Python SDK for OpenResponses Spec

Model training/serving and HELM eval server for NeurIPS evals

Showcasing the experiments done on guardrail models supported in any-guardrail during the Builder in Residence program during the summer of 2025.

No description.

No description.

A collection of Jupyter notebooks and scripts showcasing Lumigator's functionalities

A desktop application for building custom llamafiles with bundled models and configurations.

Material for the tumo workshops

JavaScript SDK for mcpd

Homebrew tap for Mozilla.ai

Hackable implementation of state-of-the-art open-source LLMs based on nanoGPT. Supports flash attention, 4-bit and 8-bit quantization, LoRA and LLaMA-Adapter fine-tuning, pre-training. Apache 2.0-licensed.

No description.

No description.

mcpd plugins Python SDK

Dummy lib to prevent typo-squatting common typo

No description.

No description.

No description.

No description.

A strict, explicit SemVer CLI with first-class prerelease support.

Identify GGUF files by their SHA256

Rust SDK for building mcpd plugins

A community driven registry service for Model Context Protocol (MCP) servers.

Protocol buffer definitions for mcpd

mcpd plugin SDK for .NET

Visual Question Answering for videos with slides

A framework for few-shot evaluation of autoregressive language models.

Agent-ready provider API wrappers with typed schemas, OAuth scope mappings, and LLM function-calling definitions.

Stateless OAuth protocol library — auth URL building, PKCE, code exchange, token refresh, revocation, scope parsing.

No description.

No description.

CLI for accessing and managing otari

No description.

No description.

Bindings to copy of SentencePiece esaxx library (fast suffix array and frequent substrings). Modified to allow dynamic MSVC linking.

Container for using any-guardrails from otari

No description.

Langchain any-llm integration

Use Hugging Face with JavaScript

No description.

Python SDK for cq — shared agent knowledge commons

No description.

Discord bot that also interacts with 0din submissions API

Adaptation of Blueprints to run on Kubeflow Pipelines (KFP).

Ray is a unified framework for scaling AI and Python applications. Ray consists of a core distributed runtime and a set of AI Libraries for accelerating ML workloads.

A toolkit to run Ray applications on Kubernetes