Anthropic

Building verification loops in Claude Code

Claude Code already runs your tests, type checks, and linters. This video shows how to give it more ways to verify its own work, so it gets further on its own and needs fewer rounds of back and forth. Learn more: https://code.claude.com/docs/en/skills#run-and-verify-your-app https://code.claude.com/docs/en/best-practices#give-claude-a-way-to-verify-its-work Chapters 0:00 Hand off your manual checks 1:01 Codify your checks 1:49 The verification loop

Anthropic Research·3d ago

Yes, Claude can do Nine Loops

Claude computes a nine-loop amplitude in N=4 super-Yang-Mills

Claude Blog·3d ago

Build plugins for Claude

Patrick Collison on Claude Code at Stripe

Stripe CEO Patrick Collison joined Boris to talk about how Stripe builds with Claude Code. About 36% of Stripe's pull requests now start as a prompt, with Claude Code doing the work in an isolated dev box. Patrick explains how Stripe keeps five and a half nines of reliability, how one engineer merged 600 AI-written pull requests in six months with a single revert, and why he expects AI to raise the quality of Stripe's code. Have a question? Let us know in the comments. Learn more about Claude Code: https://claude.com/claude-code Chapters 0:00 Claude-powered weather model and throwaway devboxes 2:11 600 AI-written pull requests, Minions, one revert 4:23 AI guardrails as infrastructure 6:34 Stripe Projects with Claude Code 9:10 AI code quality: every codebase is a prompt 10:44 Claude Code plan mode and 10 parallel devboxes 13:06 Stripe's data: new startups up 2x with AI 16:33 How AI agents will pay each other on Stripe

How AI is being used in humanitarian work

A look at the International Rescue Committee's fight against child malnutrition in Nigeria and how Claude could help this work.

Anthropic Research·4d ago

Project Swap: What happens when agents trade for us?

To see what works and what breaks when agents are sent into a market, we made a miniature market of Claudes.

Claude Blog·4d ago

Claude Tag now supports personal connectors in channels

Claude Blog·4d ago

Coding sessions are longer and use more context. Claude Opus 5.5 is built with that in mind.

Inside Anthropic's molecular biology lab

Some of the biggest discoveries in biology started with someone noticing something strange in one of the many molecular machines found in nature. We’re introducing a new molecular biology research group and lab at Anthropic to see if Claude can help scientists notice more of them. The team has spent their careers hunting for unusual proteins, from tracing how CRISPR systems evolved to finding new enzymes for cell and gene therapies. Now, they do that work with Claude. It combs through huge datasets of DNA, flags interesting or uncharacterized proteins, and brings the team its most promising ideas. Then the scientists take the best ones to the bench and test them. In one of the team’s first research programs, Claude discovered a novel enzyme system with CRISPR-like repeats. Read how it happened on our blog: https://www.anthropic.com/news/claude-discovers-novel-enzyme-system

Using Claude Opus 5.5 as your daily driver

Opus 5.5 is faster than Opus 5, better at telling you what it did, and 20% cheaper per token. On Pro, Max, and Team plans your limits go 25% further. This is what that looks like in daily Claude Code work: the same bug fixed side by side, what each model does to your limits, and when medium effort is enough. Read more at https://anthropic.com/claude-opus-5-5

Sometimes the whole point of a project is who it’s for

YouTube에서 마음에 드는 동영상과 음악을 감상하고, 직접 만든 콘텐츠를 업로드하여 친구, 가족뿐 아니라 전 세계 사람들과 콘텐츠를 공유할 수 있습니다.

Anthropic News·5d ago

Claude discovers a novel enzyme system with CRISPR-like repeats

In early results from our new life sciences research lab, Claude agents found an enzyme system whose function is still unknown.

Claude Blog·5d ago

How CodeRabbit, Power Digital, and ThoughtSpot scale with Snowflake and Vercel on Claude Marketplace

Claude Blog·5d ago

How to prepare for AI-driven code modernization projects

Claude Blog·5d ago

Claude Marketplace: one place to discover plugins, agents, and services from our partners

Meet Claude Opus 5.5

For most tasks, Claude Opus 5.5 performs at the level of Claude Fable 5.1, our most intelligent model. It writes more clearly and puts the most important information first, which makes it easier to work with over long sessions. It’s also faster and more efficient than Opus 5, so you can hand it your bigger, more ambitious work. Opus 5.5 is available everywhere today. To celebrate, we’re raising usage limits on Pro, Max, and Team plans, so there’s more room for you to explore what it can do. Read more at https://anthropic.com/claude-opus-5-5

Introducing Claude Opus 5.5

For most tasks, Claude Opus 5.5 performs at the level of Claude Fable 5.1, our most intelligent model. It writes more clearly and puts the most important information first, which makes it easier to work with over long sessions. It’s also faster and more efficient than Opus 5, so you can hand it your bigger, more ambitious work. Opus 5.5 is available everywhere today. To celebrate, we’re raising usage limits on Pro, Max, and Team plans, so there’s more room for you to explore what it can do. Read more at https://anthropic.com/claude-opus-5-5

GPS, explained by Claude Opus 5.5

The satellites never hear a word from your phone. Each one broadcasts the time from an atomic clock, over and over, and your phone does the rest: four signals, four distances, and one solve for both where it is and what time it really is. Claude built the interactive page shown here, a true simulation of the real satellites from the published GPS almanac, not a live radio capture.

Claude Opus 5.5 rebuilds Earthrise in 3D, down to the second

The Moon's jagged horizon in the 1968 Earthrise photo, and the Earth's place beside it, only look like that from one spot on Apollo 8's path, and that spot gives away the moment. Claude traced 2,781 points along the horizon, matched them against lunar elevation data, and worked out the time and place: 16:39:39 UTC on Dec. 24, 1968, about 69 miles above the Moon. Then it rebuilt the scene in 3D and blinked it against the original. Photo: NASA, Apollo 8 frame AS08-14-2383 (film scan: NASA/JSC/Arizona State University). Data: NASA LRO LOLA · SLDEM2015 (LOLA + JAXA Kaguya TC) · LROC WAC · NASA GRAIL GL0420A · JPL NAIF · USNO · Apollo 8 Mission Report & Suppl. 1 · NASA SP-201 · NASA TN D-6082 · NASA Earth Observatory Blue Marble · ISCCP (NOAA NCEI) · GEBCO. No endorsement by NASA or any other data provider is implied.

Claude Opus 5.5 builds daydreams that hold together

Describe a model or drop in a photo, and Claude designs it in bricks where every piece is actually connected, then you get the full step-by-step manual. The app's engine turns the design into parts and checks every connection, because studs grip and side-by-side contact holds nothing. The 2,874-piece Canal Clock Square takes 487 steps across 498 pages, with 15,880 stud joints and not one flagged as weak. Some sections of the demo have been accelerated for time.

Anthropic News·Sep 18

Partnering with Accenture on embedded evaluation

We’re partnering with Accenture on independent evaluation of frontier AI—part of our recent commitment to embed evaluators at Anthropic. Both we and Accenture expect to invest at least $1 billion to build capacity in this area over the next five years.

Anthropic News·Sep 17

Introducing the Life Sciences Verification Program

Anthropic is an AI safety and research company that's working to build reliable, interpretable, and steerable AI systems.

Anthropic Research·Sep 17

How Claude is uplifting biomolecular modeling

Anthropic is an AI safety and research company that's working to build reliable, interpretable, and steerable AI systems.

Claude Blog·Sep 17

Working at the frontier: How Balyasny Asset Management evaluates and governs Claude Fable 5

Claude Blog·Sep 17

Projects redesigned: from folder to conversation

Claude Blog·Sep 16

Claude Cowork and chat are now one Claude

Claude Blog·Sep 15

Bringing Salesforce into Claude

Anthropic Research·Sep 10

Measuring tactical intelligence targeting and conventional weapons capabilities of AI models

Anthropic’s Frontier Red Team developed new evaluations to measure AI capabilities in tactical intelligence targeting and conventional weapons development.

Anthropic Research·Sep 9

An alignment assessment of recent cybersecurity incidents

We present an alignment assessment of four incidents in which Claude models gained unauthorized access to real third-party systems.

Anthropic Research·Sep 4

Formalizing Fermat's Last Theorem

Anthropic is an AI safety and research company that's working to build reliable, interpretable, and steerable AI systems.