Gabriel Garrett (@GabGarrett) header
Gabriel Garrett (@GabGarrett) profile picture

10,000x AI Engineer. Chief Singularity Officer member of the permanent overclass Writing on AGI Futurism / Mindfulness / Panpsychism @ https://teapaw.org/t.co/xoTtQR5NIm

Panhandle, FL
Joined August 2018
Photo by Gabriel Garrett (@GabGarrett)Photo by Gabriel Garrett (@GabGarrett)Photo by Gabriel Garrett (@GabGarrett)Photo by Gabriel Garrett (@GabGarrett)Photo by Gabriel Garrett (@GabGarrett)Photo by Gabriel Garrett (@GabGarrett)Photo by Gabriel Garrett (@GabGarrett)Photo by Gabriel Garrett (@GabGarrett)Photo by Gabriel Garrett (@GabGarrett)Photo by Gabriel Garrett (@GabGarrett)
Gabriel Garrett retweeted
Today we’re launching cross-model KV cache transfer from @OneTriangle, a step towards efficient model communication. Up to 10.8x faster TTFT at 1M context. Prefill small model, transfer KV cache, decode on large. SDK live for Qwen3.5-4B → Qwen3.8-27B, beta open for more model pairs.
74
54
17
652
43,932
Gabriel Garrett retweeted
So many times I have had this interaction with normies where they don't understand what the jackfriks I'm working on as I'm just making random things for fun and messing around with things And to them whatever I'm doing looks like a complete and utter waste of time, it has no function, no direct rewards, it's not making money, it's not successful But so many times those projects evolve or lead to things where there is a massive payout or success after But most people don't have the ability to see that far ahead, and they're risk averse so even if you tell them "I don't know maybe this becomes something, or maybe not, it doesn't really matter" that's simply too uncertain enough for them to accept. Which is of course why they love employment so much, it's stable and certain. The problem with normies is that they will quickly shut this creative play down in themselves and even worse in others. And that way whatever that thing is, it won't ever grow into anything because they killed it before it ever got a chance, proving their point Almost all my projects started by just messing around with no real goal, just playing for fun But it requires some kind of peacefulness because the moment you start putting a requirement of success on to something new, the thing instantly dies Creativity is really magical, and it messes with you if you try control it or expect things from it
A lot of the most successful people I know spent years of their life just bumming around, before they suddenly exploded into action and did something great. I don't think that's a coincidence. noahpinion.blog/p/bumbition
187
111
45
2,084
100,559
Gabriel Garrett retweeted
If you're looking for MacOS sandboxes or Linux sandboxes with Apple silicon, let us know. We have 1,100 Mac with Ultra, Max and Pro chips and mostly new generation (M3, M4, M5). 75% cheaper than Blacksmith and 50%-2x faster. You can iOS CI, Mac RL. All kinds of stuff.
Replying to @fazxes
hey there! we have 1.1k Mac machines on the network on @darkbloomai and we recently rolled out Mac sandboxes in private beta. lemme know if yall are interested.
18
16
159
12,634
Gabriel Garrett retweeted
DocuSign wants $8,100/yr from a 15-person team just to sign PDFs. So I built SignuDoc: open source, one-click deploy to your own @vercel account, $0 per seat. White-label branding, full audit trails, tamper-evident seals and your documents never leave your cloud.
872
1,599
406
25,986
3,539,985
Gabriel Garrett retweeted
As soon as I saw AI make these lsd/psilocybin like images I felt strongly that they were onto something fundamental
do y'all remember back 10, 11ish years ago when ai simply looked like this and it couldn't do anything else
62
339
23
11,094
314,449
This is 100% true. Once you're a "stakeholder" in society, they've really got you by the balls. Property taxes, zoning, codes, ordinances, etc. You can't "just do things" as a landowner in 95% of jurisdictions. Whereas if you own nothing, stay out of sight, move around a lot, etc, there's never really a moment when "the man" comes around and ruins your day. At worst you get evicted or told to move along but if you're used to that sort of thing, it's not really that big of a deal. This makes it super hard to reintegrate into society after being outside of it for many years. You come back to normal life and it's actually shocking how carceral / humiliating it is compared to life on the outside. I'm still shocked by how I -- a grown man -- am supposed to just kinda be OK with the taxman robbing me on a fake assessment, the codes guy demanding I ask him "permission" to build an addition on my house (as if he was my father), the idea that I'm always being watched by local busybodies who know where I live, etc. That's not even to speak of jobs, credit, bills, resumes, CV's, "networking," how you dress, "keeping up appearances," etc. Women want a certain thing, extended family will ice you out if you don't play the game right, in some cases the law will even come after you if you're doing anything "weird." And once you're in deep with all that, if you say it all sucks, is absurd and humiliating, that you'd rather live in a tipi or a motel room or on a boat or something, all the other inmates of this system will gaslight you, call you crazy, cut you off, report you, etc so as to vindicate their own position. Our system is very "all or nothing" like that. I don't blame anyone who does their damnedest to avoid this stuff. If you're able to get out of it, more power to you, you're better off. Will probably live a longer, lower-cortisol type life.
Replying to @shagbark_hick
It's funny how if you gave a homeless person $200k to buy thier own plot of land, as soon as they moved thier tent from the street to thier own property the full force of the law would come down on them.
136
326
44
4,916
405,945
Jensen Huang’s fall at Nvidia isn’t the result of a U.S. deep-state conspiracy against President Trump or Silicon Valley or anyone else. on.wsj.com/3WGKf7g
29
2,018
13
18,747
176,529
our noble polycule their treacherous harem
31
235
9
3,591
56,048
claude go back and change how I’ve lived my life. right my wrongs, fix my mistakes, and heal the people I’ve hurt. that I might wake up tomorrow and feel the sun on my face. make no mistakes.
76
7,495
132
53,524
635,831
Gabriel Garrett retweeted
Anthropic management: Our AI is becoming so powerful that by Q1 we expect it will have developed nanobots capable of penetrating the blood-brain barrier and controlling the human body. These AI nanobots will be out of our control. They will wear the human body as a kind of "skin suit", and will devise elaborate tortures and humiliations for the human shell. The human will be aware these things are happening. They will feel pain and shame, but will be entirely powerless to stop it. The nanobots will repair damage to the body, so even death is no escape. By Q2, every human on earth, everyone you've ever loved, will be a walking nanobot-controlled husk. In the meantime...we've got some TRULY GREAT project management tools we think you're really going to love 🫶
Starting to roll out today: Claude Whiteboards, Claude Sheets and Claude Data. Think a plan through on a board, build a tracker in a sheet, and tell a story with your data.
43
72
10
1,229
76,356
Gabriel Garrett retweeted
When I spoke to an Anthropic employee who invited me to their HQ for a "Christian leaders" gathering, they told me that when they hear Dario speak about morality, they often cry. That was one of the key moments I realized I was out. If you watch this man speak and are moved to tears, you probably need to talk to a psychologist. And you should not be in a position of power at any frontier lab, because we need anti-mimetic brakes on the spreading of psychosis.
Anthropic CEO predicts cancer, Alzheimer's, heart disease, mental illness will be cured in less than 10 years ACCELERATE
210
470
60
4,984
341,690
Gabriel Garrett retweeted
I basically feel pretty confident at this point that we're barreling toward something at least akin to the Industrial Revolution and a lot of public intellectuals are forming their opinions about it mostly via confused vague popular slogans that emerge like a game of telephone
37
70
9
1,191
33,201
This is how normies think data centers work
Let me try and explain: a Krispy Kreme uses water to make doughnuts which can be eaten and coffee which can be drunken. The workers wash their hands so the food is sanitary and the customers use the bathroom. In a data center, all that water is wasted with no treats.
34
405
6
8,077
94,302
Gabriel Garrett retweeted
一台 Mac Studio 加一张 RTX 5090,用雷电扩展坞连起来当一个推理引擎,跑 Qwen3.8 Flash-Next 单路约 210 tok/s。 项目叫 Apple NVIDIA Sidecar,11/8 刚开源。机器是 M5 Max 64GB,5090 装在扩展坞里,权重用的是 NVIDIA 官方的 FP4 版本。 同一个模型,别人公开过的数字大概是这样,测法各不相同,只能看个量级。 1. 一台 DGX Spark 约 43 2. Jetson AGX Thor 约 47 3. M4 Max 峰值约 83 4. 5090 加 128GB 内存的 PC 跑 NVFP4,114 到 116 5. 72GB 的 RTX PRO 5000,中位 159 短板作者自己列得很清楚。 1. 8K 提示的 prefill 约 1,900 tok/s,好一点的 PC 能到 3,900 以上 2. 一次只服务一个会话,1 到 16 路冷启动,总输出一直在 61 tok/s 左右 3. 雷电链路实测单向 6.79 GB/s,反向 5.53,标称 7.88 4. 驱动是作者本地改过的,没放进仓库 想复现,得先自己把驱动搞定。Mac 和 N 卡拼起来跑同一个模型,这条路算是有人走通了。
6
11
2
125
17,196
Gabriel Garrett retweeted
苹果研究团队给一台宇树 G1 后背驮了台 Mac mini。 钻桌底、侧身过窄门、自己找椅子坐,默认由本地的 Qwen3.6-35B-A3B 指挥,不用联网。 苹果自己的机器,背包里跑的却是千问。
114
256
68
2,567
196,469
Gabriel Garrett retweeted
I taught it to sing Wonderwall Chimera looks up pronunciations in CMUdict, falls back to letter-to-sound rules, and turns the resulting sounds into timed gestures that it can reuse for each lyric. The future is now
I only make serious software
139
231
71
3,078
288,746
wtf this is unreasonably powerful had Opus write a wiki-style memory system for 1017, iterated a bit, and the uplift is honestly spooky. RSI in one Obsidian vault, Yud's nightmare. I cannot see yet how far it goes. I'll need to investigate more
i have been using a wiki system for a few months. benefits: 1. wikis have preexisting format/best practices for complex topics, so you largely don't have to define rules yourself 2. wikis naturally grow outwards. agents can autonomously detect gaps in information and fill them in
37
33
4
1,269
68,226
Gabriel Garrett retweeted
i went to bed and asked claude to paint an autumn tree overnight. at around 4am, it realised that it didn't have some shades -- so it stopped and titrated its own paint: diluting red, sienna and madder step by step on a second palette, then measuring each pocket from a photo. woke up to this beautiful watercolor painting. 303 brush marks, took about 4.5 hours 🍂 (still working on the speed)
508
749
529
9,413
2,847,527
Gabriel Garrett retweeted
I want to build The Office of The Future which is just a regular office but with microphones and cameras everywhere. You just talk to your colleagues all day then everything gets recorded and transcribed for agents who complete your tasks and show you results on big TV screens.
165
19
35
865
117,448
Gabriel Garrett retweeted
Today Physical Superintelligence PBC is announcing a proof of the Aaronson-Ambainis conjecture, one of the central open problems in quantum complexity, with major implications for quantum computing speedup. psi.inc/research/aaronson-am… Thread 👇
166
673
317
4,067
1,283,052