LLM

The FDA has cleared over 1,357 AI-powered medical devices. Every single one of them used old-school AI — pattern recognition, image classification, signal analysis. Not one ran on a large language model. That just changed. RecovryAI, a San Francisco startup fresh out of stealth, announced that the FDA granted Breakthrough Device Designation to its generative AI chatbot for post-surgical recovery. It’s the first time the agency has given this designation to anything powered by an LLM — and the implications go way beyond one startup’s product. ...

Every week, a new blog post asks: MCP or RAG? RAG or Agents? Agents or MCP? The question is wrong. These aren’t competing technologies. They’re layers — and the teams shipping the most capable AI systems in 2026 are stacking all three. Here’s what each layer actually does and why you probably need more than one. MCP: One Plug to Rule Them All LLMs are brilliant and trapped. They can reason about your calendar but can’t read it. They can plan a database migration but can’t execute it. They need hands — and before MCP, giving them hands was a mess. ...

Anthropic just cannibalized its own flagship model — and that’s exactly the point. Claude Sonnet 4.6, released today, delivers near-Opus performance at $3/$15 per million tokens. That’s one-fifth the cost of the model it’s chasing. In benchmark after benchmark, the mid-tier model is breathing down its big brother’s neck. In some cases, it’s already ahead. This isn’t a minor version bump. It’s a price-performance earthquake. Two Models, Twelve Days, One Message Anthropic dropped Opus 4.6 on February 5th with “Agent Teams” — the ability to spin up multiple AI agents that coordinate in parallel. Think less “chatbot” and more “autonomous project team.” A million-token context window. PowerPoint integration. The works. ...