winstonp

6 ms·

1.
by winstonp 14d ago
It's the perfect model for open-source work because it's gonna end up in the training data anyway
2.
by winstonp 15d ago
I think it's overbloating my computer
3.
by winstonp 28d ago
It's a habit in specific countries, China is one in my experience.
4.
by winstonp 1mo ago
Cursor was 60B. The 10B number was the breakup fee if the deal fell through.
5.
by winstonp 2mo ago
16 tokens / s is not nothing.
6.
by winstonp 2mo ago
Apple isn't counting on their model to be a frontier coding and cowork model. Gemini is perfectly fine for the tasks that new Siri is supposed to be doing.
7.
by winstonp 2mo ago
That is not the Grok CLI being discussed. That's an open source, third party CLI. https://x.ai/cli is the official Grok CLI being discussed, and it is not open source.
8.
by winstonp 2mo ago
The avg coding session has hundreds or thousands of tool calls. Even a 5% failure rate noticeably notches up token use and cost. See Gemini.
9.
by winstonp 2mo ago
Doubt. Tried it, Grok 4.5 is leaps and bounds ahead of Gemini for real work. And token costs vs. raw compute+electric cost is unit profitable too.
10.
by winstonp 2mo ago
While I mostly agree with your post, I do want to point out one thing: > Or he could release a model trained largely by existing open weights models. Which without some huge breakthrough probably has no chance of surpassing them, so is p
11.
by winstonp 3mo ago
same happened to Opus 4.7
12.
by winstonp 3mo ago
the British are notoriously sensitive to heat. They'll call 30 Celsius weather a heat wave.
13.
by winstonp 3mo ago
do we think he found it on craigslist
14.
by winstonp 3mo ago
Gemini pretty clearly has the best underlying model, and the worst RL and post-training of the lot.
15.
by winstonp 3mo ago
It is ridiculous to be building infrastructure with a seventy five year lifespan under an assumption that may never come in its entire life.
16.
by winstonp 3mo ago
Car apps, beside Tesla, are universally awful to use. Even Tesla's is not beyond reproach (app size is massive, for one), but at least it doesn't make me want to poke my eyes out. Apple should make a "Cars" app that&#x27
17.
by winstonp 3mo ago
They are absolutely clueless about how to talk to this administration.
18.
by winstonp 3mo ago
CC TUI is often a culprit of memory leaks, image-pasting is a nightmare, particularly over SSH, flickering all over the UI, among others.
19.
by winstonp 3mo ago
You can! You just have to pay API prices.
20.
by winstonp 3mo ago
google models are still very unreliable at actually calling the tools you want it to call.
21.
by winstonp 3mo ago
I don't believe Anthropic trains on Trainium, only serves models on it.
22.
by winstonp 3mo ago
Claude Code TUI is garbage. There's nothing worth protecting in there.
23.
by winstonp 3mo ago
I do think Anthropic nailed their naming down much better than OpenAI
24.
by winstonp 3mo ago
OpenAI's training is better suited to developing models that don't have these tendencies
25.
by winstonp 4mo ago
I've never had a smelly Mac, and I've owned maybe 10 different ones across personal and various work laptops. And 90% of devs I've ever met have used Macs and none of them smell either, so it's zero out of maybe 200+ in
26.
by winstonp 4mo ago
Google Search has been awful for the last couple of years. Good riddance to SEO.
27.
by winstonp 4mo ago
Same. Overuse of the word "genuinely" is another tell.
28.
by winstonp 5mo ago
Which apps have you seen ask for someone to setup a local LLM? Can't recall having ever seen one
29.
by winstonp 7mo ago
The new Studio Displays have more powerful chips than the Neo.
30.
by winstonp 7mo ago
Agree. Claude tends to produce better design, but from a system understanding and architecture perspective Codex is the far better model

More ›