Tag: Claude
-
Claude Spent 21 Hours and Found a New Piece of Biology
Anthropic’s scientists set a few hundred Claude agents loose on a database of roughly 200,000 enzymes, gave them 21 hours, and asked them to find something nobody had described before. They did. It’s called ART, short for array associated reverse transcriptases. It shows up in bacteriophages, the viruses that infect bacteria, and it looks like…
-
Claude Just Stopped Asking You to Pick a Lane
Anthropic just did something small that actually matters. They took Claude Cowork, the tool built for handing off bigger, longer jobs, and folded it straight into regular Claude chat. No more picking a mode before you even start typing. If you have used Claude at all, you know the moment. You open the app and…
-
Google Just Let Its Own Engineers Use Anthropic’s AI
For a long time, Google had a simple rule for its own people: build with Gemini, full stop. Employees weren’t allowed to touch outside coding tools like Claude Code or OpenAI’s Codex. If you wanted a taste of Claude, you basically had to be on a short list. That rule just quietly loosened. Google opened…
-
Anthropic Has a Smarter Model. You Won’t Get It
Anthropic just told the world it built something stronger than its own best public model, then told the world it is not letting anyone touch it. The company calls it Model 2. It beats Mythos 5, Anthropic’s current flagship, and there are no plans to ever release it. This came out of Anthropic’s August 2026…
-
Claude Just Learned How to Learn From You
Anthropic released a new feature for Claude this week called ‘Teach Claude a Skill,’ and it is deceptively simple but potentially significant. You record your screen, narrate what you are doing, and Claude watches and learns. Then next time you need the same thing done, you just ask. This is rolling out to Claude Pro,…
-
Microsoft Just Handed You Choice
Microsoft shipped 40 updates to Copilot this month. The one that matters: Claude is now an option inside Copilot Chat. This is not small. For months, the model wars felt like watching walled gardens argue through marketing. Now? A knowledge worker in Microsoft 365 can match the tool to the task. Need to shred a…
-
Anthropic Just Gave You the Mirror Test
Anthropic launched Reflect this week, and it’s doing something we haven’t seen before: tracking how you actually work with AI. It’s a dashboard that lives in Claude. You can see what topics you spent time on, your peak working hours, how much you’re using Claude, patterns in what you ask for. It generates a monthly…
-
Claude Sonnet 5 Just Added Caching and the Math Actually Works
Anthropic shipped prompt caching on Claude Sonnet 5 and the economics are not a gimmick. This is actually useful right now, not in theory. Here is the math. You write a cached prompt at $2 per million input tokens. Every time you read from that cache within 1 hour, you pay 10 cents per million…
-
Anthropic Just Made Fable 5 Expensive on Purpose
Fable 5 went from subscription-included to credits-only starting July 8. Pro subscribers now pay $10 per million input tokens and $50 per million output tokens. That is the most expensive pricing Anthropic has ever put on a publicly available model. But here is the catch: Anthropic got backlash and extended free access through July 12.…
-
Claude Enterprise Just Got Cost Controls That Work
Anthropic released admin controls for Claude Enterprise that let IT and finance teams see exactly where their Claude bill is coming from and set spending limits before they get surprised. The problem being solved is real. Agentic AI tools can burn through tokens fast. A query that used to cost cents now costs dollars when…