@beffjezos Too similar to the SBF/FTX playbook back in the day
English
Ernesto Ramirez
276 posts
Ernesto Ramirez
@ernestor_eth
techno-optimist | engineer | building @succinctlabs
San Francisco Katılım Mayıs 2022
706 Takip Edilen438 Takipçiler
Anthropic is strategically trying to burn the AI vibes to the ground so people overregulate and the game gets frozen while they're in the lead
Claude@claudeai
There’s hope in hard questions.
English
Scare everyone about AI, then pose as the only ones humble enough to be trusted.
Claude@claudeai
There’s hope in hard questions.
English
@fakedev9999 It's good, but significantly nerfed. It downgrades to opus too often for random shit. 5.6 sol is pretty cracked
English
I don’t have subscription since July 1. Am I missing out? Or is Fable 5 still a trash?
Claude@claudeai
We're extending Claude Fable 5 access on all paid plans, as well as keeping Claude Code’s weekly rate limits 50% higher, through July 19.
English
@gakonst What do you think is the right balance between working through Centaur and devs using their individual sub/seat? Bc Centaur costs significantly more per token and iteration time seems slower for larger scope tickets.
English
Plz take the extra effort to own your data and your infrastructure, assign 1 person on your team to set up Centaur over Claude Tag and let me know how it goes - docs in reply.
Claude Tag is awesome, I've spent a few reps using it to understand where it does better than our OSS alternative (just memory right now which we'll have next week and have intentionally delayed because we didn't think it was worth it) but it's just hard to risk your entire company's brain atrophying if that were to be taken away from you.
And soon we'll allow you to swap out frontier models for finetuned OSS ones on each channel (working on the eval rn) and then your cost is going to be in the low tens of thousands per year vs 10-100x that, plus you owning it all. That's the catch the forces above us don't want you to know.
Georgios Konstantopoulos@gakonst
It seems extremely important for companies to be able to own their data and later on their models by finetuning open source models on their data. Products like Claude Tag are obviously against that, as the incentive is to deeply integrate and max out the data you can get. We built an OSS tool which empowers companies to get the "virtual collaborator" today on Slack/Teams etc. with existing hosted models, self-host it, capture all their own data, and then whenever the team feels like it, they can eject into sovereignty! centaur.run
English
I’m convinced @sama calls for an emergency meeting everytime this guy posts a video
Husk@huskirl
Is this good music?
English
Very impressive. Now put these on a @BostonDynamics Atlas.
1X@1x_tech
NEO’s Hands An API to the Physical World
English
Imagine @GavinNewsom shows up at your BJJ gym for a trial class. That'd be fun. Maybe not for him, but definitely for everyone else.
English
@advaith @iHateMyBtch You should sign up for the next love island season
English
Among the other things AIs will get better at is recognizing AI-generated text. Presumably writing generated by this year's models will be easy to catch in a few years. Prediction: This will cause huge scandals in academia, as published papers are later exposed as AI-generated.
English
2002: I'll google it
2015: I'll uber over
2026: I'll GPT it (???). I'll claude it (???)
Why haven't we picked the dominant verb for AI yet?
English
Last night in SF I hosted the world's largest Peptide Polycule
Jeff Tang@jefftangx
I'm hosting a Peptide Tasting Party on Tuesday in SF We have a 5-course meal: 1. Natural peptide powders from whole food ingredients 2. Bioregulator capsules 3. Sublingual drops 4. GHK-Cu topical cream 5. BPC oral film Come early. While supplies last. > Tag a friend who should go > RT for awareness > DM me for the invite link
English
@garrytan We'll eventually see some form of Berlin Wall in New York or California if enough people leave these states.
English
We aren’t doing this trash in San Francisco because we saw what these kinds of ruthless politicians do to the citizens
Asian elders killed
Public education dismantled
If you start counting their bodies and talking about their abdication of duty, the hard left can’t win
Lit News Network@litnewsnet
NEWS: Three Democratic Socialist candidates won primaries in New York, cementing a big shift in the Democratic party. Here is a summary of each of the candidates and their radical stances: Claire Valdez (NY-7): Grant citizenship & voting rights to illegals, use taxpayer funds for all transgender treatments, & eliminate private health insurance. Darializa Avila Chevalier (NY-13): Abolish prisons, abolish ICE, abolish borders, defund the police, and “all deportations are wrong”, including for violent criminals. In a now-deleted tweet, she said “I forgot to get napkins so I just wiped my hand on the American flag behind me.” Brad Lander (NY-10): Abolish ICE, forgive all student loans (~$2 trillion), expand the Supreme Court.
English
@advaith @clementetv_ @fakedev9999 Someday this is what coding by hand will feel like, at least to the hand coding enthusiasts
English
@AlexFinn The question is when is it worth investing in hardware to run a model like this, purely for sovereignty and personal use? Hard to justify spending $15-30k on hardware right now. Most would rather pay for tokens and forego privacy.
English
As a sidenote, this is a MASSIVE model
You will not be running this on a Mac Mini. In fact, you will need quite a few high end GPUs
But here's the thing: hardware will get better. Models will be made more efficient. This is the worst it will ever be
Within the next few years this will 100% be able to run on widely available consumer hardware. I'd bet ANYTHING on it
The fact that a model who's weights you can download right now is just as good as Opus 4.8 is a feat in its own
English
I was wrong
I've been saying for months that open source AI models are 6 months behind frontier
They caught up. GLM 5.2 is as good as Opus 4.8
This changes everything. If you run GLM 5.2 locally no government can take it away. You become sovereign
And even if you run through APIs, its a fraction of the cost
The battlefield is different now. If open source is as good as frontier, and people have cheaper alternatives, governments can't be as quick to regulate. It will destroy the frontier AI labs
All of this is such a massive win for the people
If you are not paying attention to local models yet, you are making a tremendous mistake
Z.ai@Zai_org
Introducing GLM-5.2: Frontier Intelligence, Open Weights - Significant improvements in coding and agentic tasks - Strong long-horizon capabilities with a 1M context window - Two levels of reasoning effort: GLM-5.2 (max) pushes the limits, while GLM-5.2 (high) strikes a strong balance between performance and token efficiency - MIT-licensed open weights - Same API pricing as GLM-5.1 Tech Blog: z.ai/blog/glm-5.2 Weights: huggingface.co/zai-org/GLM-5.2 API: docs.z.ai/guides/llm/glm… Coding Plan: z.ai/subscribe Chat: chat.z.ai
English
