🚀 We’ve just launched our new Mozilla.ai website! We’ve given our website a full makeover, making it easier than ever to explore our work and connect with what we’re building. 👀 Check it out and explore what’s coming next: www.mozilla.ai

We're excited to announce that our ML Systems Engineer Raz Besaleli, and Head of AI Engineering Irina Vidal Migallón will speak at the DataFest Tbilisi (Nov 27–29)! Raz on lessons from systems engineering for resilient AI, and Irina on key risks when working with AI. More info: www.datafest.ge

Working with multiple platforms & LLMs, but some models only appear on certain providers? Try any-llm Model Finder! See connected providers via API keys & search models across all of them. It's built on top of our any-llm library. Check out our repo and GIF: github.com/mozilla-ai/a...

Join us for an interactive preview of Mozilla.ai’s Agent Platform! We’re inviting HR, CX, Sales & Marketing experts for a hands-on session to test and build a working agent that addresses a task of your choice. No technical skills required. 👉 Register: https://luma.com/bj15wy0k

Guardrails are vital for AI safety, but hard to choose. Each has its own logic & definition of unsafe. That's why we built any-guardrail, a unified interface to test and switch guardrails seamlessly, making AI safer, more reliable, and easier to build. Read our post: blog.mozilla.ai/introducing-...

Last Friday, our PM Stefan French spoke at AI_dev Summit about @mozilla.ai's open-source projects: Blueprints and any-suite libraries. We aim to help devs adopt open-source AI with confidence, innovate faster, and avoid vendor lock-in. Check out our slides: static.sched.com/hosted_files...

What if AI agents ran entirely in your browser? Baris Guler takes that idea seriously in our first-ever guest post on the Mozilla.ai blog: 🧱 WebLLM + WASM + WebWorkers 💻 Rust, Go, Python, JS 🔒 Fully local. No API calls. Read the post here: blog.mozilla.ai/3w-for-in-br...

🚀 New releases! any-llm v0.15.0: List available models Local demo chat app Better async support Now supports llama.cpp & Perplexity any-agent v1.5.0: any-llm is now default for tinyagent framework Check them out: github.com/mozilla-ai/any-agent github.com/mozilla-ai/any-llm

We just released mcpd: requirements.txt for agentic systems. It simplifies running MCP servers with one config that works across local, CI, and prod. Props to Peter + Alejandro, and the team at Mozilla.ai repo: github.com/mozilla-ai/m... blogpost: blog.mozilla.ai/meet-mcpd-re...

🚀 New releases for our any-suite tools! any-agent v1.2.1: API consolidation, better small model support, test suite extension, A2A streaming. any-llm v0.12.0: New providers, OpenAI Responses API, reasoning output, improved stability. 🔗 github.com/mozilla-ai/a... 🔗 github.com/mozilla-ai/a...

At @mozilla.ai, we tested OpenAI’s new gpt-oss reasoning model and hit a challenge: providers handle reasoning output differently, needing to diverge from OpenAI Completion API spec. any-llm fixes this by standardizing outputs, so you can compare providers easily. 🔗 blog.mozilla.ai/standardized...

Watch @daavoo.bsky.social talk at @mlopscommunity.bsky.social virtual conference, and explore how you can unlock new ways to understand agent behavior with structured traces that capture the agents' full chain of thought, tool calls, and model interactions. 🎥 www.youtube.com/watch?v=yC3X...

🚀 any-llm v0.5.0 is out! - Added support for new providers: OpenRouter, PortKey & LMStudio - Supports embedding models inference via API. More flexibility, better integration. Thanks to our community for the feedback! 🙌 Repo: github.com/mozilla-ai/a... Docs: mozilla-ai.github.io/any-llm/prov...

🚀 any-llm v0.3.0 is live! ✅ Streaming now supported for AWS, Google & Mistral ✅ Expanded test coverage for better stability Another step toward a simpler, unified interface that works seamlessly across providers. 🔗 Check out our repo: github.com/mozilla-ai/a... Try it out & share feedback!

Our ML Engineer Nathan Brake joined the Firefox podcast Outside the Fox to talk about AI security: trending scams, identity protection, and how @mozilla.ai contributes to building reliable and transparent AI agent systems with tools like any-agent. 🎥 Watch here: www.youtube.com/watch?v=ZDnD...

Announcing any-llm, a Python library that lets you switch LLM providers with a single config change. Key features: ✅ Uses official SDKs ✅ No proxies ✅ Supports OpenAI, Mistral, Anthropic & more ✅ OpenAI format responses Read more on our blog post: blog.mozilla.ai/introducing-... Repo & docs below

any-agent v0.25.0 is live! • Callback module • Batch processing support for parallel tasks • Structured output improvements • Push notifications • Support for Mistral API • LLM-readable docs • New tracing hooks, test modules, and cleaner config experience for a2a ⬇️ Links in next post

New any-agent release (v0.24.0) is out. ▪️ AgentJudge returns full eval traces ▪️ Serve agents via MCP ▪️ Use agents as tools ▪️ Cost + token tracking ▪️ A2A protocol compatibility We're making agents more composable + observable. ⬇️ Links in next post

🌐 New blueprint: AI agents that run in your browser via Webassembly & Pyodide. Just open an HTML file and go. They work with OpenAI API compatible models or local models via Ollama/vLLM. Take a look : github.com/mozilla-ai/wasm-agents-blueprint

🎯 New Mozilla.ai Blueprint: Speech-to-text alignment Tired of Whisper butchering your name or technical terms? With Michael Hansen from the @openhomefoundation.org , we built a tool to align any Whisper model to YOUR words. Take a look : blueprints.mozilla.ai/all-blueprin...

Ready to build AI agents? Here’s the first Any-Agent Cookbooks! Cookbook 1: Build your first AI agent Cookbook 2: Create an interactive booking assistant with MCP Explore the cookbooks: - mozilla-ai.github.io/any-agent/co... - mozilla-ai.github.io/any-agent/co...

New Blueprint: Test and compare AI agents frameworks using a real-world use case: finding the best surf spot! 🔄 Implement any-agent to switch between agent frameworks ⚙️ Test different agents & models combo 📊 Evaluate performance Demo: huggingface.co/spaces/mozil... Code: github.com/mozilla-ai/s...

Generative AI is powerful but unpredictable. Small prompt changes can cause inconsistent or biased results. We tested DSPy for prompt optimization, and built a prompt saliency tool to understand LLM behavior. 📖 Read more: blog.mozilla.ai/smarter-prom...

🧠 Choosing the right LLM is hard. Our latest post explores how LLM Routing helps devs intelligently direct queries between models based on accuracy, speed, or cost . Read more on our blog : blog.mozilla.ai/the-challeng...

🎥 Watch the Mozilla.ai x @eleutherai.bsky.social webinar recording! See experts demo the Audio Transcription & Document Conversion Blueprints, including setup, best practices, and tips. Recording: mozilla.hosted.panopto.com/Panopto/Page... Explore more: blueprints.mozilla.ai/all-blueprints

We're launching our first-ever Builders in Residence Program for practical, application-driven AI researchers (PhDs, recent grads, industry). Fully remote (EU, US, CA), ~3-month fellowship starting June 2025. Learn more & Apply here: job-boards.greenhouse.io/mozillaai/jo...

🧪 Testing AI ≠ testing software. From hallucinations to prompt injections and data leaks, LLMs introduce new risks. Traditional testing won’t cut it. We unpack the challenges (and tools) in our new blog post: 👉 blog.mozilla.ai/what-do-you-... #AI #LLM #SoftwareTesting #MozillaAI #TrustworthyAI

Synthetic speech is getting too realistic. Worried about misinformation? Train your own detection model with our Blueprint by @mozilla.ai and UncovAI. Lightweight, easy to deploy, superior results. Get started: github.com/mozilla-ai/f...

🎙️ Our ML engineer @aittalam.bsky.social explains "Why Thinking Small Is The Future of AI" on @kathleenchu.bsky.social 's Progress Reimagined podcast. www.youtube.com/watch?v=HDM8...

🚀 New Lumigator Updates Are Live! ✅ Provided better translation support with Opus-MT models and G-Eval integration. ✅ Expanded evaluator capabilities by integrating DeepEval's Ollama. ✅ Refreshed the Experiments UI with new features and improvements. 👉 Try it out: github.com/mozilla-ai/l...

Are AI apps really tested? Most devs rely on manual checks. So we built AIssert: a lightweight tool to test LLM-powered apps built on @haselrovi.bsky.social ✅ No Python ✅ No LLM config ✅ Plug into any CI/CD Repo: github.com/mozilla-ai/a... Blog: blog.mozilla.ai/aissert-test... #MozillaAI #AItesting

🚀 New! Mozilla x EleutherAI Blueprints for easier & open dataset creation Two powerful tools for developers: 1️⃣ Audio Transcription using privacy-focused Whisper models 2️⃣ Document Conversion to Markdown for building open-text datasets Available now on the Blueprints hub: blueprints.mozilla.ai

Join our user research! Mozilla.ai is talking to devs of all experience levels building with AI. 📅 30-min chat over the next 2 weeks 💬 Sign up: forms.gle/9vyjDs7EfLXM... 📝 Or take our anonymous survey: forms.gle/JBYFAeLjwB4T... We'd love to hear from you! #MozillaAI #UserResearch #AI

Register now for GLOSAIC – The Global Open Source AI Conference 🌍🧠 Stefan French will be presenting Mozilla.ai Blueprints, our open-source starter code for real-world AI use cases 📅 1:10 PM ET on Wednesday 16th of April. Don’t miss it! Register here: glosaic.org

Meet Any-Agent, a Python library that lets you build and switch across agent frameworks. Try it: github.com/mozilla-ai/a... We built it because we were struggling. Now we’re sharing. 🔹 Unified interface 🔹 Logging with OpenInference 🔹 Built-in eval Blog: blog.mozilla.ai/introducing-...

Big night in Lisbon! 🇵🇹 We co-hosted an incredible AI Agents meetup with #Lumiera at #Cloudflare's #Lisbon office. From surf spot recs to production #MLOps, our speakers shared real stories from the frontlines. 🎥 Missed it? Watch here: cloudflare.zoom.us/rec/share/DC... (🔑 a*NBk84X)

Mozilla.ai is going to university! This Thursday, Mario Cariñana will speak at Universitat Politècnica de València (UPV) on open-source AI, our work at Mozilla.ai, and building trustworthy AI systems. ¡Nos vemos el jueves!

Lisbon AI friends! Join us tomorrow at Cloudflare for a Mozilla.ai x Lumiera x #MLOps meetup on AI agents. 🔹 Real-world agent use 🔹 Deployment lessons 🔹 Talks from Mozilla.ai, Tripadvisor & Unbabel 🍻 Drinks, snacks and good convos included! lu.ma/hmytvzm3?tk=...

Mozilla.ai is in Málaga for our team workweek! Big ideas, great energy, and finally some in-person face time for our remote team <3 #MozillaAI #AI #Workweek #RemoteTeam

💡 Can local LLMs deliver strong quality while keeping your data private? We used Lumigator to benchmark 4-bit models (deepseek-r1:14b, gemma3:12b) via Ollama, Llamafile & vLLM on WMT24++. Local is catching up fast. Privacy, cost & offline use matter. 🔗 blog.mozilla.ai/evaluating-l...

🚀 We’re excited to share our new Blueprints Hub (blueprints.mozilla.ai). Explore all our Blueprints and get access to instant demos, video walkthroughs, technical insights, and real-world use cases. 💪Join us and be part of a growing open-source AI community!

🚀 AI Translation & Beyond: Lumigator’s Latest Innovations! 🌍 Join us March 25 | 4 PM GMT on Mozilla AI Discord (discord.com/invite/SAKRU...) for a live demo & Q&A on multilingual AI, model deployment & translation evaluation. #AI #Translation #Lumigator #LLMs

📱🔒 Build your own timeline algorithm! BYOTA Blueprint creates a local feed for your goals, not engagement metrics. Powered by Mastodon.py, llamafile & marimo. Re-rank fediverse posts locally: github.com/mozilla-ai/b...

What’s the smallest model that gets the job done? 🐤 We benchmarked DeepSeek R1 for zero-shot clinical summarization using Lumigator 🐊 to test efficiency, accuracy, and cost trade-offs. Read more: blog.mozilla.ai/benchmarking...

Stockholm AI x Mozilla.ai brought great discussions on open & trustworthy AI last week! 🔹 Lumigator -> we tested the new translation eval and LLM-as-a-judge feature. 🔹 Blueprints -> received key feedback on onboarding & docs. 📅 Next up: lu.ma/hmytvzm3

The Federated Fine-tuning Blueprint uses Flower AI to let multiple data owners collaboratively fine-tune AI models while keeping data private. Example: Qwen2-0.5B-Instruct on Alpaca-GPT4. Get Started now: github.com/mozilla-ai/f...

We put DeepSeek V3 to the test using Lumigator, how does it compare to BART on summarization? Read more: blog.mozilla.ai/evaluating-d... 📌 Gains in METEOR & ROUGE 📌 But BERTScore suggests smaller returns vs. compute cost 📌 Highlights the need for better evaluation metrics

📍🗺️ Passionate about maps and want to train your own computer vision models? The OpenStreetMap AI Helper Blueprint uses YOLOv11 & SAM2 to make @OpenStreetMap contributions more efficient while keeping humans in the loop. Help us find more swimming pools. Get started here : github.com/mozilla-ai/o...

Blueprints Deep Dive: Get hands-on with our open source AI building blocks. Join us Tuesday Mar 4 @ 1PM EST via Discord to explore Structured Q&A and Speech-to-Text Finetuning Blueprints. discord.gg/D2RzGn73?eve... Save the date and bring your questions.

github.com/mozilla-ai/s... Need to fine-tune speech-to-text (STT) models? Our latest Blueprint helps you adapt STT models for specific accents, languages, or domains, using your own local data or the commonvoice.mozilla.org datasets.