I usually avoid viral stuff, but the chatgpt instagram โfor youโ page prompt is consistent with my values of radical candor/vulnerability and interest in using Ai as a thought companion. So โฆ
https://t.co/hk8Yazidfq
Mythos invented its own language, then switched back to English to talk to humans
(AI safety researchers have been warning of this "Neuralese" risk for years. If AIs stop reasoning in English, we can't monitor their thoughts, which means we can't detect scheming.)
Thank god for xaiโs unused gpus โฆ this to me feels even bigger than every release before it. Set it and forget it makes fable universally valuable. Anthropic rev is gonna moon even harder.
Introducing Claude Fable 5: a Mythos-class model that weโve made safe for general use.
Its capabilities exceed those of any model weโve ever made generally available.
I think pg is genuinely โcuriousโ but the likeliest answer is fairly obvious imo โฆ binaries are intrinsically engaging; they form the basis for the dramatic arts ffs
Why are AI-generated replies so often framed as an opposition between two things? Is there a popular bot that does this? Are such tweets known to be especially engaging? Or is this just what an AI does by default when you ask it to write a tweet?
Claude code hanesss should obv check for latest cc capabilities. Insane every user needs to do this manually each time or build their own skill (like me) to call it.
(I think /whats-new is built it in but maybe I built it a lot time ago idk)
been asking others at Anthropic how they stay in the loop with Claude and fully understand the work being done
this is one of my favorites from Suzanne:
Eli Lilly has done it.
They've gone and made what seems to be a powerful, permanent gene therapy for LDL cholesterol.
That means they'll be able to effectively prevent most heart disease with a single infusion!
Launching our new paper on arXiv: we trained the largest multilingual food model ever built.
4.1M recipes. 7 languages. 1,790 ingredients. 300 dimensions.
All of human cooking compressed into 2 megabytes.
Proud of this one.
We shipped a new Institutional Holdings API.
This data comes from 13F filings, which is a nightmare to parse at scale.
You must normalize tons of data.
Funds will report "BERKSHIRE HATHAWAY", "BRK", "BRK.A", and the CUSIP across different positions.
Resolving these to a canonical security requires a reference โ CUSIP โ ticker map that is correct both today and historically.
You need to track ticker changes, splits, and M&A activity, which breaks the majority of vibe coded approaches.
Also, there is a ton of data: your DB table will easily exceed 1B rows if done right.
We solved it @findatasets. It's now 1 API call.
@thisischristina i just think presenting stocks to children in anything but the warren buffet mindset of owning shares of businesses (that is stock prices aren't important) is really unhealthy and likely to encourage gambling and money oriented mindsets (thankfully i overcame my greed)
@thisischristina the rest is thoughtfully designed so you should know that what's "above the fold" dominates mindshare.
i fell in love with investing/stock market and started reading the wsj cover to cover at 13, so i'm no anti-capitalist.
Weโve agreed to a partnership with @SpaceX that will substantially increase our compute capacity.
This, along with our other recent compute deals, means that weโve been able to increase our usage limits for Claude Code and the Claude API.
Our agreement with @SpaceX means we will use all the compute capacity at their Colossus 1 data center.
This will give us over 300 megawatts of additional capacity to deploy within the month.
Effective today, we are:
1) Doubling Claude Codeโs 5-hour rate limits for Pro, Max, and Team plans;
2) Removing the peak hours limit reduction on Claude Code for Pro and Max plans; and
3) Substantially raising our API rate limits for Opus models.
nice. anthropic is effectively lowering prices after jacking them up substantially during and after the demand spike they experienced over the last couple months
Weโve agreed to a partnership with @SpaceX that will substantially increase our compute capacity.
This, along with our other recent compute deals, means that weโve been able to increase our usage limits for Claude Code and the Claude API.
Boris Cherny says internally Anthropic uses the same models as everyone else, with some Mythos, which will eventually ship as a descendant to the public
"there's no manually written code anywhere at the company"
Internally, Claudes talk to each other all day over Slack, coding in loops, resolving unknowns across teams
New for financial services: ready-to-run Claude agent templates for building pitches, conducting valuation reviews, closing the books at month-end, and more.
Install them as plugins in Cowork and Claude Code, or use our cookbooks to run them in production as Managed Agents.
@Squee451 Codex is good at code. Which opus is also good at. 5.5 is decent at charts and data, almost nothing else ime.
Doesnโt even follow normal instructions 5.2 and 5.3 did.