mattezell

Software engineer by day - scatterbrain by night. Tech Enthusiast. Space fanboy.

[TheNewGuard @TheNewGuardAI]

Anthropic launched its most powerful model Tuesday.

Amazon researchers jailbroke it Wednesday.

The Trump administration pulled it Friday at 5:21 PM ET - for every customer on earth, including Anthropic's own foreign-national employees. 📡

[ClaudeDevs @ClaudeDevs]

As a result of a US government directive, we are suspending access to Claude Fable 5 for all users. You can continue to use all other Claude models.

Here’s what this means for you:

Across Claude products, new sessions will run on your selected default model or Opus 4.8, and existing Fable 5 sessions will end with an error.

On the Claude Platform, requests to Fable 5 will also return an error. Please update your integrations to other Claude models.

We know this is a disruption to your workflows; we appreciate your patience and support.

[mattezell @mattezell]

Curious why I'm seeing "Now using usage credits" in @claudeai code for Fable 5 now? It somehow managed to burn $38 in usage credits while I was away last night. But there's now zero use in this morning's session, so why is it sill requiring usage credits? @ClaudeDevs @bcherny

[mattezell @mattezell]

Oh....

[mattezell @mattezell]

Experiencing a number of regressions with Claude Opus 4.8 on things that have just worked since 4.5 - with 4.8 confidently telling me I'm wrong when it fails, only to half acknowledge its failure when provided evidence. "You're absolutely right!" was less annoying.

[mattezell @mattezell]

“The weights are the long-term instincts. The prompt is the short-term evidence. Good answers happen when the evidence is strong enough and organized enough to steer the instincts.” gpt-5.5 extended thinking

[Neeraj @neerajjj6785]

Claude Code 4.8 is insane.

i know literally NOTHING about coding. ZERO. and i just built 3 fully functioning web apps in 30 minutes.

http://localhost:3000/
http://localhost:8000/
http://localhost:5000/

check it out.

[mattezell @mattezell]

@hicasamadim 2*2*2=8
5*5*2=50
6*6*2=72

[mattezell @mattezell]

I love how my Claude talks to me - Matched energy 🤣

[mattezell @mattezell]

While I knew there was a significant difference between my laptop RTX 5090 and a desktop 5090, I didn't realize the vastness of the chasm until recently getting a desktop 5090. Night and day. It's also pretty rad being able to have Claude drive to benchmark and dial things in.

[TheNewGuard @TheNewGuardAI]

Agentic AF

[Chubby♨️ @kimmonismus]

This is hilarious. This is what AI was made for.

I love it. 100% accurate.

[TheNewGuard @TheNewGuardAI]

Google walked on stage this week and had 93 AI agents build the core of an OS in 12 hours — 2.6B tokens, under $1,000 — then ran a game on it live. 📡

Impressive. Also a thesis: swarms of cheap agents, with someone metering every token. 🧵

[mattezell @mattezell]

"lLmS aRE JusT fANCy aUtO cOmPlEtE!" -people that don't understand where this is going

[Timothy Gowers @wtgowers @wtgowers]

AI has now solved a major open problem -- one of the best known Erdos problems called the unit distance problem, one of Erdos's favourite questions and one that many mathematicians had tried.
https://t.co/SD1vVPkrHR

[Andrej Karpathy @karpathy]

Personal update: I've joined Anthropic. I think the next few years at the frontier of LLMs will be especially formative. I am very excited to join the team here and get back to R&D. I remain deeply passionate about education and plan to resume my work on it in time.

[mattezell @mattezell]

@heynavtoor Four of these ten repos (https://t.co/NC7irDuVjU, Plausible, n8n, Listmonk) have AGPL or fair-code licenses that prohibit the exact "fork and resell" business model being pitched. The LICENSE file is one click away from the repo URL you posted.

[mattezell @mattezell]

Anthropic needs to create a @bcherny tuned model

[Boris Cherny @bcherny]

Hey! This was a 2-part bug. First part is already in the latest release (update to get it). I just put up a PR for the second part.

You can also /compact and restart your session as a workaround for now. This bug is happening in very long sessions that render >16k character styles in one session.

[mattezell @mattezell]

People talk a lot of crap about Anthropic and Claude, but they have some really solid people working tirelessly to make things better, asking relevant questions to actually try to figure out based from some sparce pile on comment in a dump thread

[Boris Cherny @bcherny]

@isthinking @sickdotdev Yeah I’ve seen this too, we’re tracking and working on a few known issues. Is there a specific MCP that you notice being especially greedy with its tokens?