TensorFlow.js

Introducing Veo 3.1 and Veo 3.1 Fast 🚀- now with richer audio, better prompt understanding, and enhanced realism! Give it a try to craft the look of your scene, create longer seamless shots, and control the shot from start to finish! 🏁 🔊 Featuring enhanced realism and audio…

October 17th is just around the corner! This is your final call to register for the in-person #WebAI summit in the Bay Area. Don't miss out on a full day of talks and demos on client-side AI. 🎟️ Grab your spot before it's too late → goo.gle/WebAISummit2025

What does an agentic internet look like? 🤔 @jason_mayes walked us through how to create and run client-side AI agents in JS, powered by smaller LLMs, to perform useful work faster on existing websites using Web AI. Link to our YouTube video in the reply and subscribe to get…

Come join us in the Bay Area for Google's #WebAI summit, where you'll hear from leaders at @HuggingFace, @Arm, @Google and more. If you can make it in person, sign up here → goo.gle/WebAISummit2025. If not, no need to RSVP—videos will be on @YouTube soon!

Customize with ease, deploy with speed. Adapt and run Gemma 3 270M locally for your own tasks: > Fine-tune the model in minutes with Google Colab > Reduce the model footprint to under 300MB > Run it on-device using MediaPipe and LiteRT Learn more: developers.googleblog.com/en/own-your-ai…

Jason Mayes (@jason_mayes) is coming to #CascadiaJS to show us how to create and run client-side AI agents in JS, powered by smaller LLMs, to perform useful work faster on existing websites using Web AI. ⚡ Get a glimpse of the future—an agentic internet. 🤖 More info 👇

Okay this is insane... WebGPU-accelerated semantic video tracking, powered by DINOv3 and Transformers.js! 🤯 This will revolutionize AI-powered video editors... which can now run 100% locally in your browser, no server inference required (costs $0)! 😍 Who's building this?

Real-time conversational models🔥, continuous thought machines🧠, and a world tour reaching ~10K developers 👩‍💻 The August #WebAI Monthly is HERE! ✨ Read & sub right now to be the most interesting person in the room at the next AI meetup you attend 😉 Massive shout out to the…

Photo from the stage! Full room and lots of great questions around client side #WebAI today - thank you for joining me at #WCEU2025 - I look forward to seeing what you all create with this technology when combined with #WordPress - hybrid of cloud and local running in browser…

Thank you to Barry Pollard who got some shots of me in action on stage at #WCEU2025 today ✨ And of course thank you to all of the developers in attendance learning about the possibilities of #WebAI within the #Wordpress ecosystem 🇨🇭 Next up Berlin for Google IO connect and then…

See how new CSS features, AI in Chrome DevTools, and multimodal AI using Gemini Nano can empower you to build a more powerful and modern web ⬇️ Senior Tech Writer @AlexandraScript recaps the top web announcements from #GoogleIO! youtu.be/GSVe6zguiao

😁Hollywood demos that make you smile (literally), 💻new client-side multimodal models, and Transformers.js set to power new Firefox AI abilities 🔥 The May #WebAI Monthly is HERE! ✨ Read & sub right now to be the most interesting person in the room at the next AI meetup you…

Visual programming frameworks like Visual Blocks can speed up ML pipeline building, but novice users can still struggle with setting them up. Read how InstructPipe, a research prototype for ML pipeline design, helps users build workflows more easily →goo.gle/4lB4pav

In a language detector API, which result do you prefer? und always last: [(it: 0.69), (es: 0.15), (und: 0.16)] descending probability: [(it: 0.69), (und: 0.16), (es: 0.15)] (und = unknown. Actual API shape is { language, confidence }.) Discuss on github.com/webmachinelear….

We are so excited to have a special @jsmonthlylondon where you can learn all about: RSVP now - guild.host/events/jsmonth… Tools for #Accessibility by the @foolproof accessibility guru @fjordaan and All about #WebAI from @jason_mayes Lead #WebAI @Google Register now on…

🔊Sound on! 🍿 It appears I just recreated that 1 friend at movie nights who loves to talk witty remarks in scary scenes... Somehow using TTS in a 🇬🇧 British accent makes it even better. Hybrid AI demo using #WebAI in JS for TTS + multimodal to test the water. If you want the…

Yo, #WebAI Devs, just wanted to let you know that $200k up for grabs from mother Google for supporting solid #WebML work in your existing activities, such as conversion of models to work in the browser for #JS folk to use, or creating tools and utilities that can benefit the…

A few people were asking me how to cache the #MediaPipe Gemma 2 #WebAI model in the browser so that the 2nd page load is fast (as 2.5 GB download). Turns out it was a little tricky due to upper limits of how large a blob can be in browser cache, so I made a library to help you…

🤖🎶 What if you could reuse my #WebAI #Agent code + apply it to entirely different use case? Turns out you can! Made this Web AI @Spotify #DJ in 1 day, running entirely client side in @googlechrome! Reuse my code for any industry or API - including yours - to perform useful…

I'm all for research creating top Gen AI models like #DeepSeek at lower costs and such, but the real power and money to be made will be the developers who use such models to solve real problems. It's time to get creative, like I did in this #WebAI demo that uses Gemma 2 entirely…