larping as a software dev
Joined June 2025
- Tweets4.8K
- Following224
- Followers58
- Likes15.5K
Pinned Tweet
i was tired with how google calander worked so like i made my own and its less bad
and far faster to add events
its heavily wip, theres some rough edges but check it out :)
github.com/vvxcrystl/lmnaide
crystl retweeted
"i asked grok" "i asked chatgpt" well i held council with the memories of my collected ancestry and they commanded me to destroy thinking machines
crystl retweeted
So DeepSeek and Kimi learned from GPT-6 Astra. I genuinely do not give a fuck.
Everyone in business copies. Always did.
Industrial espionage is older than every AI lab on this planet.
Every American tech giant copied someone on the way up.
The question is not who learned from whom.
The question is what these models did to the market.
DeepSeek does a browser task for $0.22.
GPT-6 Astra charges $10.77 for the same task.
That is competition. Prices fall because of it, and we all win.
Meanwhile look at how the US side competes.
Federal agencies ordered to remove DeepSeek.
Bills written wide enough to cover open models.
Fear campaigns about Chinese weights running on your own computer.
The Chinese labs do not answer any of it.
No mud. No crying on X. They just work quietly and ship weights.
I know which side behaves like a competitor.
teapawt.net/kimmonismus/status/211…
crystl retweeted
sam: local inference is a terrible allocation of scarce resources
“a lot of people want local inference, and somehow they assume that the problem is solved now because we have open-source models, we're going to be able to run them on our computers, and so we're not going to need all the stuff in the data centre
but from a global allocation perspective, actually it's quite terrible to have everybody, like, have a bunch of the world's limited supply of DRAM sitting in their own computer and being used five percent of the time
you want maximum utilisation in the cloud if you're trying to deal with this very supply-constrained environment
somehow many people's intuition is the opposite of that”
crystl retweeted
High-key getting tired of the "Fuck AI, it's all bad" or "fuck generative AI" and so on so forth.
I'm not for, nor against, AI as a whole; but everyone crying about it need to get over it... I understand peoples' viewpoints, especially for artists, musicians and so on, I do-
crystl retweeted
Qwen4, coming soon…
CTO quote was “next week”
Xudong Guo - Research Scientist, Qwen, Token Foundry
Live from Amsterdam
The existence of "Deepseek v5 Open" implies that one of them is gonna be closed weight.
are we gonna see Closed weight deepseek models?? Yikes.
crystl retweeted
>buy 3 DGX Sparks
>"now I'll never be short on compute"
>inference needs all 3
>research needs all 3
>mfw I'm short on compute
