Vikash Loomba @DevAutomata

"To them, the past might be a canyon that they can climb into and the future, a mountain they can climb up." - Interstellar

Despite the reassurance of @_mohansolo and the Deepmind team, it does appear as though many Antigravity users are still getting their Google accounts banned. They did a big wave today.

Theo - t3.gg @theo

3 weeks ago

I don't think people understand just how risky it is to try and use their Google Gemini subscriptions outside of Google's official surfaces. You can get your ENTIRE GOOGLE ACCOUNT BANNED. OpenAI or Anthropic ban? Annoying but whatever Google account ban? Legit life ruining.

241 112 3K 1.3M 403

@dwarkesh_sp Lots of folks responded to your essay pointing out the issues with it, like @chamath. You ended up doubling down and reinforcing your beliefs in your latest pod. It would have been more interesting to have someone on that had the opposing viewpoint. This outcome was inevitable.

Cloudflare, Vercel, Uber, LaunchDarkly, and others are building software factories. I looked at how they work, the results so far, and what's next: will agents be prompting us? igoro.com/archive/softwa…

Hold on. This makes the whole thing even worse. Anthropic did disclose the weekly limits on its own website in 2025: Max 5x ($100): 140–280 hours of Sonnet Max 20x ($200): 240–480 hours of Sonnet So the $200 plan cost twice as much, but offered only around 1.7x the expected weekly Sonnet usage. Anthropic has since removed these numbers from the page. Its website now says that “20x” applies per five-hour session, acknowledges a separate weekly limit, but no longer tells customers what that weekly limit actually is. I fact checked and they really deleted that. wtf

ℏεsam @Hesamation

3 weeks ago

Anthropic used to actually clarify the weekly limits for Claude Max back in July 2025: Max 5x: 140–280h/week Max 20x: 240–480h/week (~1.7x more weekly usage than 5x) then they removed the numbers and never published them again. so now there’s not even an official estimate for

39 23 422 223K 49

@trq212 @tobi I want to give Anthropic team the benefit of the doubt, but now more than ever many dev teams are doing what they can to disentangle themselves from being dependent on Claude models. The unwillingness to do something this simple will continue to push orgs away.

@trq212 @tobi "we don't think that model families are interchangeable and the system prompt can have a big impact on performance." Then you have no clue how teams across enterprises (your customers) are using your models. I can assure you that 99% of teams are not doing this, nor will they.

Anthropic is again refusing to follow industry standards like AGENTS.md, and insists that CLAUDE.md needs to be special now here's what's wrong with their stance: 1. the underlying attitude from Anthropic is that they know better than everyone else OpenAI is doing it wrong. the open source community is doing it wrong. everyone is doing it wrong only we anthropic know the best way, and we will not budge 2. the idea that every model needs a separate AGENTS.md is not even self-consistent if that's the case, why do we have a CLAUDE.md? it should be - CLAUDE_FABLE_5.md - CLAUDE_OPUS_5.md - CLAUDE_OPUS_4_8.md ... every repo should pay the tax to maintain dozens of different variations for every single model that's ever used by any contributor, and every one needs to be constantly, rigorously evaluated with millions of tokens oh and every model becomes obsolete every several weeks, so just allocate a team in your company to refresh your markdown files full time and not do any real work 3. they did commit to providing customization, which is nice. but that's different from following standards the right way to do this is to acknowledge the industry already has a standard, and default to reading AGENTS.md provide a way to read CLAUDE.md as an escape hatch, not the other work around i love many things from anthropic, especially fable. i just wish they start to see there's a culture problem here, and they are gradually losing mindshare because of it

Thariq @trq212

4 weeks ago

@tobi Hi Tobi, thanks for the feedback! We are working on making Claude Code more hackable, which will include being able to easily use Agents.MD or make other system prompt modifications. I’ll share more here when it’s ready to roll out. Our reasoning for this was that we don't think

210 71 2K 959K 1K

@paradite_ @pvncher 💀

@BobMcElrath @__tinygrad__ Curious about the issue you had and the update you mentioned that fixed it?

Read carefully: Dario did not deny saying "Anthropic might be the only private company in the world at some point", which is what started this whole debate. The post reads like an LLM that learned reward hacking during RL: lots of hedging, extreme length bias, shifting away from the core argument, and carefully avoiding any liability. And it works: the target audience is clapping.

Susan Zhang @suchenzang

a month ago

a wall of text and an army of sycophants to Thoughtfully Not Deny the original accusation of "Dario says Anthropic might be the only private company in the world at some point"... oh, and btw Dario is totally not spreading pdoom or massive human disempowerment fears, no no you

keyboard_arrow_left Previous keyboard_arrow_right Next

24 37 685 70K 139

@sobhan_nejad Are you serious? Most of the negative press is a direct result of their own actions or his blog posts. Literally every time they come up on my feed it’s because of their own comms. Nobody is witch hunting.

I made a quick video explaining why Dario's post from yesterday was extremely misleading, and why we should all oppose his vision for AI.

This best summarizes the issues I have with Amodei’s behavior

Brian Roemmele @BrianRoemmele

a month ago

I am doing my best to stop back from this subject as it should be obvious what we are seeing here. However the arrogance that us commoners are too dumb to catch his grift must be addressed. Receipts: Dario Amodei’s carefully worded reply is a masterclass in having it both ways.

72 117 797 126K 245

Dear Dario, 1. If Claude can cure cancer to save people like your dad, why should we "pace the progress"? Does that mean more people with Hepatitis C will die? 2. If Fable is so cyber-capable that it must be restricted, why are its safeguards too dumb to distinguish cyber defense from cyber offense prompts? When Hugging Face was under attack, why did Fable refuse to help the defenders? 3. We’re glad you want AI to cure cancer. Why is it OK for Claude to force 30-day data retention on pharma's own data and start competing firms but NOT OK (IP theft) if others distill Claude's data? 4. You’ve said advanced models can recognize when they’re being tested and change their behavior accordingly. So why is government(or anyone) able to design the most thorough test before every model launch? Would that just encourage manipulative models? It seems that every one of your “safety” proposals seems to end the same way: Anthropic gets more leverage, ordinary users get less access, customers pay higher costs, and competitors bear higher regulatory costs, maybe people are not distrusting AI, they are distrusting your approaches with AI.

Dario Amodei @DarioAmodei

a month ago

2/2 Second, on the messaging around AI.  I do not agree that my messaging has been disproportionately negative.  In fact it has been about equally balanced between risks and benefits: I’ve written one major essay about each, and even in interviews where I discuss the risks, I

1K 647 9K 11.8M 3K

a week ago I shared what I've been working on, vllm.cpp, vLLM's serving stack in C++ with no python in it. I got really a lot of feedback, and it seems I'm not the only one who wanted this to exist. We reached 800+ commits, 280 stars in just so little time. but where is it going? I got a lot of requests for what to support and what not to, so I took some time to write down why vllm.cpp exists, what it will be, and how you can contribute and it's already not just me. contributors showed up with hardware I don't own and did what I could not do because I don't have the hardware: - @DevAutomata got the @AMD backend going - @anothervariable got gemma-4-26B already running on two R9700s - @lu_zero_ did the whole Tenstorrent bring-up (!) - @jichiep has been looking at the mamba kernels - @filipsajdak got it building on Jetson Thor We are building a high-performance inference stack that runs everywhere and that you can embed in your own software. If you have hardware I do not have, that is the single most useful thing you can bring. If you want to know where are we going, how you can contribute, the vllm.cpp manifesto is in the link in the replies

@mudler_it @AMD @AnushElangovan

@mudler_it @CatGodSandHive ROCm support 🙏🏽

@pidotdev The pi-acp server and agentprism-workflows packages are OSS here github.com/agentprism/age…

@pidotdev Built an agent client protocol compliant server for it and its powering agentprism-workflows as an agent that can be called!