Product @ Anthropic

Joined April 2007
96 Photos and videos
I had a great time at CwC Japan showing off some cool features of the Claude Platform: Prompt Caching, Tool Search, Programmatic Tool use, compaction, and Advisor. All get better with Fable 5:-) thanks @rodhowarth for being a great demo partner
1
7
250
Brad Abrams retweeted
I’d heard from some that they felt model progress since December was incremental, things were slowing down. But Anthropic was only publicly offering their second best model. This model card kind of reminds me of that scene in The Princess Bride where Inigo Montoya says “I’m not left handed”.
Claude Mythos and Fable 5 Benchmarks Anthropic is SOTA everywhere
3
2
91
19,718
Brad Abrams retweeted
Claude Fable 5 is our first generally available Mythos-class model. It ships with new safety classifiers that may flag certain prompts in dual-use domains like cyber and bio. We've added fallbacks: a refused request retries on Claude Opus 4.8 instead of dead-ending.
199
335
5,141
395,139
As a powerful model, Fable does hit classifiers a bit more frequently, so we developed an easy solution with server side fallbacks: platform.claude.com/docs/en/…
1
3
532
Fable has been a huge change for how I work. I am able to delegate high level long running, complex tasks and with minimal oversight. I can't wait to see what you build! anthropic.com/claude/fable
2
9
401
Brad Abrams retweeted
We just published internal data on how much of Claude's development is already being done by Claude: - Over 80% of all code merged into our codebase is now written by Claude - It's been months since many researchers at Anthropic hand-wrote code - The typical Anthropic engineer ships 8x as much code as they did in 2024 - On the most open-ended engineering tasks, Claude's success rate jumped from ~26% to 76% in 6 months - When research sessions went off-track, Claude proposed a better next step than the human took 64% of the time We're not at recursive self-improvement yet, but it could come sooner than most expect. I highly recommend reading the full blog post.
Our internal data shows Claude is accelerating AI development—a possible path to recursive self-improvement, or AI autonomously building a more capable successor. It’s happening faster than we thought, and the implications deserve greater attention. anthropic.com/institute/recu…
189
216
3,265
420,543
And... claude knows how to use ant CLI very well, give it a try!
We’ve added a CLI for Claude Platform to make every API endpoint runnable from your terminal. Call the Messages API, stand up Claude Managed Agents, pipe results straight into your shell. The ant CLI is well understood by coding agents (Claude Code) using the claude-api skill.
1
3
1,776
Small new in the Claude API: set max_tokens on the advisor tool to cap its output. Predictable cost and latency for agents that consult an advisor every step, and stop_reason: "max_tokens" tells you when the cap was hit so your agent can adapt. platform.claude.com/docs/en/…
1
3
247
Brad Abrams retweeted
there's a ton of demand right now for excellent infra talent, definitely outpacing supply im hiring engineers who have tackled megascale, multicloud product infra to work on the Claude Platform. help me find them 🙂
17
15
158
16,306
The easy way to migrate to 4.8... takes all our best practices and applies them to your code in one step...
Replying to @ClaudeDevs
We’ve updated Claude Code's built-in claude-api skill migration guidance for 4.8. Run "/claude-api migrate" to update your model strings and suggest prompt improvements that are tuned for Opus 4.8.
7
645
Opus 4.8 is out: better judgment, catches its own mistakes. My favorite API addition: mid-conversation system messages. Append {"role":"system","content":"new rules"} mid-task. Instructions update without breaking the prompt cache, and reminders keep long runs on track.
May 28
Introducing Claude Opus 4.8: it builds on Opus 4.7 with sharper judgment, more honesty about its own progress, and the ability to work independently for longer than its predecessors. Available today at the same price.
1
14
989
Lots of good tips here
1/ Some things I've learned recently running coding agents on large-scale projects. Most of this contradicts advice from 6 months ago!
1
6
3,434
Brad Abrams retweeted
Claude Console now gives diagnostics on prompt cache misses. it will attribute cache misses to messages, system prompt, tool, or model changes.
4
4
48
5,028
Welcome to the @StainlessAPI team. So happy to be working even more closely together ❤️
today, the @StainlessAPI team is joining @AnthropicAI ! we're bringing together experts in SDKs, CLIs, and MCP servers with the creators of MCP to make agents built on the Claude Platform even more connected & powerful. anthropic.com/news/anthropic…
23
3,593
Brad Abrams retweeted
May 11
The Claude Platform on AWS is now generally available. AWS customers get the full set of Claude API features, with AWS authentication, billing, and commitment retirement.
380
784
9,570
2,156,667
My @ClaudeDevs Code with Claude talk is up: Getting more out of the Claude Platform. I built an agent for superheroes — turns out the real villain was the token bill. Caching, context management, and the Advisor took it down. 🦸 youtube.com/watch?v=7oO37GRh…
1
1
3
568
Had a blast on stage at Code with Claude with @mariorod1 — how @github ships Claude to millions of devs across Copilot and pushes quality up cost down: caching, harnesses, and the new Advisor pattern. Thanks Mario 🙏 @ClaudeDevs youtube.com/watch?v=y5TmF_6o…
1
5
599
Brad Abrams retweeted
come say hi at Code With Claude tomorrow! ill be giving a talk w @jess__yan on some new things we’re excited to share. join the livestream here: claude.com/code-with-claude
11
8
213
11,049