look ma, no spans!

This is super cool, great demo! 🫶

Crush those bytes!

As a side-effect of the @gradio-hf.bsky.social hackathon, we may have just accidentally created the largest open source collection of MCP servers. If you are curious about what can be done with MCP you should really check them out, it’s pretty wild: huggingface.co/spaces?filte...

We have just launched the @gradio-hf.bsky.social hackathon but it isn’t too late to enter! The hackathon has over 1 million dollars in credits and prizes! For all of my svelte friends, there is something here for you too! One of the tracks is ‘custom components’, which are all written in svelte!

Building real-time WebRTC and Websocket applications is very difficult to get right in Python. Until now - Introducing FastRTC, the realtime communication library for Python ⚡️ huggingface.co/blog/fastrtc

introducing Gradio Sketch 🖌️ A no-code mode to start building and deploying machine learning apps. Install Gradio and just type: 𝚐𝚛𝚊𝚍𝚒𝚘 𝚜𝚔𝚎𝚝𝚌𝚑 in your terminal.

With the latest version of Gradio, you can launch a multi-page web in 6 readable lines of Python I actually think this might be the easiest way to start a multi-page web server in any framework, in ANY language

We're holding office hours on Discord tomorrow to share recent releases & what else is on the @gradio-hf.bsky.social roadmap! Would love to have you all join us and answer any questions you might have! discord.com/events/87954...

We did it. Finally added support. To everyone who asked for it. Especially users coming from Streamlit. You can finally have your sidebars. 𝚙𝚒𝚙 𝚒𝚗𝚜𝚝𝚊𝚕𝚕 --𝚞𝚙𝚐𝚛𝚊𝚍𝚎 𝚐𝚛𝚊𝚍𝚒𝚘

Nice

No waayy, I've been waiting for this feature!!! Now I can finally say to my friends that they can download my apps😎

🤯 Wild! Gradio apps now available on your DESKTOP or PHONE 🤯 Thanks to progressive web applications, you can now *download* gradio apps. Simply set `.launch(pwa=True)` and your gradio app will be downloadable -- its already the default on 🤗 Spaces, so download any Space now!

💪🔥 Now it is possible to Build with ComfyUI, and Deploy with Gradio on 🤗 Spaces!

Hi folks we just did a release Gradio 5.12 ahead of the weekend. The big thing in this release is a fresh UI for agent flows in gr.Chatbot 💥

a programming language whose errors are optimized for easy correction by an ai agent e.g. if an unsupported kwarg is passed into a function, it should respond with the list of supported kwargs as well

Don’t stop talking about Palestine

When is this monstrosity going to end?

There's no excuse to use closed models anymore

Just going through the list. Lots of great stuff coming to Gradio in January!

If you're looking for a little holiday project, its now dead simple to make Slack apps, discord apps, and intercom-style website widgets (even multimodal ones) with @Gradio Full tutorial: www.gradio.app/main/guides/...

Editable chatbots... coming to Gradio!

I'll get straight to the point. We trained 2 new models. Like BERT, but modern. ModernBERT. Not some hypey GenAI thing, but a proper workhorse model, for retrieval, classification, etc. Real practical stuff. It's much faster, more accurate, longer context, and more useful. 🧵

Introducing Moonshine Web: real-time speech recognition running 100% locally in your browser! 🚀 Faster and more accurate than Whisper 🔒 Privacy-focused (no data leaves your device) ⚡️ WebGPU accelerated (w/ WASM fallback) 🔥 Powered by ONNX Runtime Web and Transformers.js Demo + source code below! 👇

We just released Gradio 5.9.0, with a much more beautiful and capable ChatInterface. This is a big release, let us know what you think!

New handle!

🚀 Exciting AI Breakthrough: Microsoft Introduces Phi-4, a awesome 14B LLM Key highlights: - 14B is suddenly smoll - Outperforms larger models in math competition problems - Available on Azure AI Foundry (HF next week) - Surprise, they used synthetic data Full details: https://buff.ly/3ZNFZBY

It’s been 5 years since I have attended a conference in person!

Resonates...

Working on making the Gradio Chat UI much cleaner. Stay tuned!

Yeah I was talking to my friends on the inside and they're saying medicine and bread is short past 2 weeks. My friend is building a startup over there. He uses solar cells to charge his laptop. Super resilient guys

Have been working on something new on #ScientificAgent for the last few months and super excited to share a sneak peak of what it looks like ⭐️ Feel free to DM me if you're interested in trying it out and providing feedback!

Small models are really a big deal. The Qwen 2.5-1.5B parameter model, which came out in November, is the most downloaded model from Hugging Face ever

IN BROWSER

This was my last week at OpenAI after ~3.5 years. I worked on API safety and then transitioned to Policy Research where I worked closely with @milesbrundage.bsky.social on AI governance, frontier policy issues, and AGI readiness. As is tradition, here’s the message I shared on Slack:

Pretty remarkable for an open model that can run in your Macbook!

My friends, please write more.

QwQ is the #1 model on Hugging Face right now!

This demo of structured data extraction running on an LLM that executes entirely in the browser (Chrome only for the moment since it uses WebGPU) is amazing My notes here: simonwillison.net/2024/Nov/29/...

BREAKING: The Oxford Union overwhelmingly votes 278-59 today to declare Israel an apartheid state committing genocide. 🇵🇸🇵🇸🇵🇸🇵🇸🇵🇸🇵🇸🇵🇸🇵🇸🇵🇸

The most realistic reason to be pro open source AI is to reduce concentration of power.

A dataset of 1 million or 2 million Bluesky posts is completely irrelevant to training large language models. The primary usecase for the datasets that people are losing their shit over isn't ChatGPT, it's social science research and developing systems that improve Bluesky.

An AI researcher that wants to stop big tech owning everything was permabanned here for releasing a dataset of 2M posts. A librarian received death threats for a 1M post dataset. The EU funded the creation of this dataset of 235M posts months ago, and… nothing? zenodo.org/records/1108...

Thank you @bsky.app team for correcting the mistake. Glad to be back!

Did you know that 99% of email today is spam? Your inbox isn’t 99% spam because AI is used to filter it. The same 99% will happen here too, but if AI researchers continue to get perma-banned for making available the datasets needed to filter it, it’s going to make this platform unusable.

We just deployed Qwen/QwQ-32B-Preview on HuggingChat! It's Qwen's latest experimental reasoning model. It's super interesting to see the reasoning steps, and with really impressive results too. Feel free to try it out here: huggingface.co/chat/models/... I'd love to get your feedback on it!