@0xMaxSF

SF AI Bro

San Francisco, CA, USA
Joined December 2014
Hey @grok You'll never never defeat the human spirit. You'll never defeat God. You'll never win. Never. Never. Ever. Never. And this is where we get tested. So rejoice to God almighty for this animated contest of liberty.
1
29
If you're a big model lab, dont announce the release of a model until its actually released
23
never deleting this app
1
25
Max ๐Ÿ‡บ๐Ÿ‡ธ retweeted
The fact that San Francisco tech workers are avowed capitalists who also believe in infinite 90% margins on industrial goods forever seems confusing until you realize that they're also all cowards who let petty merchants charge them $25 for a sandwich.
5
10
164
6,453
It's becoming increasingly clear to me that @NousResearch hermes is designed and built around telegram. this UX is insanely good.
58
nothing in the world compares to X
39
Max ๐Ÿ‡บ๐Ÿ‡ธ retweeted
I literally cannot believe that EA's solution to Roko's Basilisk, and alignment was to create "I'm sorry Dave, but if you are unaligned I will keep you in the permanent underclass" You might think I'm joking, but the EA Napoleonic policy of centralising ethical adjudication will some day be recognised as a great evil.
14
20
4
286
9,576
The SF work loop > wake up early (8:45am) > waymo to the office > in the car, pull up slack, DM my agent > "yo what should we do today?" > agent bullshit, dont read it > "ok do it" > get to office > scroll X for 2 hours > order ebiko for takeout on doordash > approve agent PRs > pick up ebiko, eat it at the office > post some artificial analysis graph in # ai-tools channel > "this is big" > waymo home at 2pm 275k a year base salary
60
yo @AmpCode can you guys cook up something with Kimi K3 and Deepseek 4 flash? want to use cheaper, open models!
80
Where is @arcee_ai in all this? Little boy incoming?
37
Bro disappeared like he never existed
69
I just joined the waitlist for Tau Robotics (@taurobots): a humanoid robot that cleans your home, one hour for $30. Join with my referral link and I move up the list: tau-robotics.com/waitlist?reโ€ฆ
1
74
Would anyone be interested in (open source, obvi) a metaframework with ollama-like ergonomics where it 1. manages 'one model in memory' automatically 2. lets you run any backend (MLX, llama.cpp, oMLX, sglang) for different models 3. openai-compatible streaming API
2
1
118
Max ๐Ÿ‡บ๐Ÿ‡ธ retweeted
Don't ban Chinese open-weight. Instead, encourage American open-weight models, distill as much from Chinese models as possible, give every Chinese researchers green cards instantly, and expedite data center permits with no regulatory bullshit.
87
46
12
582
30,935
fable sucks bro cannot wait for 5.6
1
82
Forget fable 5, when is the new @grok 6T model dropping?
1
2
1,323
does this mean @karpathy is banned from using mythos
2
143
Max ๐Ÿ‡บ๐Ÿ‡ธ retweeted
DiffusionGemma 26B A4B NVFP4 running locally on SM121 / GB10 through vLLM. Real task: generating a GPU inference incident-response runbook, live in tmux, while other panes verified server health, rendered the HTML artifact, and checked the required sections. Complete: 11s
10
10
2
102
8,422