Why running local models isn’t just a technical question. @GiadaPistilli in @techreview

- Demo: - Blog post from ColQwen team:

AudioRAG is becoming real! Just built a demo with ColQwen-Omni that does semantic search on raw audio, no transcription needed. Drop in a podcast, ask your question, and it finds the exact chunks where it happens. You can also get a written answer. What’s exciting: it skips

5️⃣ Custom tool descriptions! Use api_description to make your tools stand out and tell users exactly what they do. 📝💬.

4️⃣ Authentication is a breeze—just type headers as gr.Header and Gradio handles the rest. Secure and simple. 🛡️🔑.

3️⃣ Instantly turn OpenAPI specs into MCP tools. With just one line of code, connect your backend API to any LLM—no more tedious manual mapping. 🧩💡.

2️⃣ Real-time progress updates! Gradio now streams progress notifications to your MCP client, so you’re never left guessing where things stand. ⏳🔔.

1️⃣ Seamless file uploads! No more manual file hosting—Gradio’s new “File Upload” MCP server lets agents upload files directly to your app. Less hassle, more flow. 📁⚡️.

🚀 Building AI-powered apps with @Gradio MCP servers just got a major glow-up! Here are 5 new features making dev life easier and smoother 👇

⚠️ AI isn’t just a tool—it’s also supercharging medical misinformation. From fake studies to hallucinated citations, the risks are real when governments and charlatans lean on unchecked AI.

🤖 On the farm, AI is running the show: self-driving tractors, robot fruit pickers, and drones are making “hands-off” farming real. The future of food? More tech, less sweat.

🧠 Nearly 3/4 of teens are turning to AI companions for advice, friendship, and comfort. The upside? New connections. The risk? Losing touch with real-world relationships. Wild times for growing up.

💰 Ex-OpenAI stars just raised a jaw-dropping $2B for Thinking Machines Lab. The AI talent wars are heating up, and the next breakthrough might come from this stealthy new contender.

🤝 Nvidia’s CEO Jensen Huang is walking a tightrope in Beijing, balancing US-China tech rivalry while keeping Nvidia at the heart of the AI revolution. The stakes? Trillions and global influence.

Today’s reads: Nvidia’s CEO navigating US-China tensions, a $2B AI upstart, and why so many teens are confiding in chatbots. Here’s what I found most interesting 👇.

👇 Check out the full post @Xianbao_QIAN and I wrote to explore what Kimi K2 can do, how to try it, and why it matters for the future of open-source LLMs

Most importantly, their move from closed to open source signals a bigger shift in China’s AI scene — following Baidu’s pivot. But as Yang puts it: “Users are the only real leaderboard.”.

The secret weapon: Muon optimizer, a new training method that doubles efficiency, cuts memory in half, and ran 15.5T tokens with zero failures. Big implications.

A trillion-parameter model that’s surprisingly efficient: Kimi K2 uses a mixture-of-experts architecture (32B active params per inference) and dominates on coding/math benchmarks.

A CEO who thinks from the end:Yang Zhilin (31) previously worked at Meta AI, Google Brain, and Carnegie Mellon. His vision? Nothing less than AGI — still a rare ambition among Chinese AI labs.