nateb2022

6 ms·

31.
by nateb2022 1mo ago
I agree there, I think if we differentiate between first and second order AI companies (companies producing models vs applying those models to everyday tasks) I do think a lot of the hype comes from the 2nd order companies, since they neces
32.
by nateb2022 1mo ago
There's a lot of that type of slop on LinkedIn (with accompanying slop comments) but they're not taken anywhere as seriously as the more thoughtful blog posts like this Also in technical circles, we're pretty aware of the mar
33.
by nateb2022 1mo ago
Most of the alarmist writing I see about AI displacing workers or suddenly and drastically changing the economy seem to be written by people who are pretty removed from technology or the actual businesses in which AI is allegedly replacing
34.
by nateb2022 1mo ago
Your 2x-4x cheaper claim is really hard to verify without transparent pricing for input/cache/output. Seems like you don't want to be easily compared to other providers, like on OpenRouter? Edit: some back of the napkin math
35.
by nateb2022 1mo ago
Tangentially this makes me wonder how large Opus really is. Perhaps Opus is a lot smaller than most of the 1T+ assumptions, just a lot more post-training/ finetuning on a 300-400B sized MoE model.
36.
by nateb2022 1mo ago
The $299 introductory price is not bad at all; for comparison, a Raspberry Pi 5 with 16GB RAM runs you ~$350 (Adafruit). And this Ventuno Q includes 40TOPS that the Pi doesn't come with by default, the Pi AI HAT+ 2 costs another ~$220
37.
by nateb2022 2mo ago
pyreqwest has been delightful so far: https://github.com/MarkusSintonen/pyreqwest It's been a pleasure to use, has a httpx compatibility layer, and it's fast ( https://github.com/MarkusSintonen
38.
by nateb2022 2mo ago
I think the weighted average takes into account all providers (some DS4 flash providers are 'premium' providers and offering higher speeds for higher pricing) and these are tilting the scale
39.
by nateb2022 2mo ago
updated thanks
40.
by nateb2022 2mo ago
Slightly more expensive than the (post-price hike) DS4 flash pricing, but in the ballpark. https://openrouter.ai/compare/deepseek/deepseek-v4-flash-073...
41.
by nateb2022 2mo ago
Curious why they're using 3.6-27B and not 3.8-27B which is competitive with Opus 4.6 ( https://huggingface.co/Qwen/Qwen3.8-27B )
42.
by nateb2022 2mo ago
With an astonishing 160 commits on July 18th to this one repository (see https://github.com/taylannuhogluofficial-png ), and no other activity in any other repository, I would be cautious of this seemingly benign offering. A
43.
by nateb2022 2mo ago
I'm guessing Oura has less legal money than both of those, which maybe influenced the complainant's decision to target Oura specifically, in this docket
44.
by nateb2022 2mo ago
The author's AI applied a thin layer of varnish to what is more or less handwaving. I think more people need to take an undergraduate level statistics course; the conclusions made are completely unsupported by the random trials. Weighi
45.
by nateb2022 2mo ago
A human would not have applied any of these terms to the submission, let alone their amalgamation.
46.
by nateb2022 2mo ago
"1960s-noir-pulp-fiction-rugged-detective" is incredibly obtuse, do you see any of these applying to the submission?
47.
by nateb2022 2mo ago
This is tangential to the commenter's inability to explain their choice of words which is atypical of a human who would have went with "dense", and suggests a 3T parameter model reaching for "1960s-noir-pulp-fiction-rugg
48.
by nateb2022 2mo ago
Could you expand on how it's pulp-fictitious? Or noir? Or "rugged" detective? Why the "rugged"? I've never heard of the word "hardboiled" referring to anything but eggs.
49.
by nateb2022 2mo ago
> 1960s-noir-pulp-fiction-rugged-detective style Straight out of another LLM. Please name some books in this style, since it seems to be on the tip of your tongue.
50.
by nateb2022 2mo ago
> ds pro and flash are fast but not something id actually use for important things, unlike sol, fable and maybe 3.7 here and there As someone who's used Gemini 3.7 Flash (Google sub mostly for the storage) and DS4 Flash a lot (~6B t
51.
by nateb2022 2mo ago
Well OpenRouter went from ~$5M in mid 2025 to ~$140M about a year later, so although 50x might be compared to current revenue, they probably factored in some growth models to approximate its value over time.
52.
by nateb2022 2mo ago
To bring people back to earth for a bit, two observations. 1. Current battery technology wouldn't enable continuous recording in such a form factor for longer than ~1hr. 2. Were continuous recording to be enabled, the earbud case would
53.
by nateb2022 2mo ago
On macOS (zig 0.16.0) a compile with zig build -Doptimize=ReleaseSmall yields a 5.8M binary for me.
54.
10x: The federal government's own venture studio(10x.gsa.gov)
2 points by nateb2022 2mo ago|0 comments
55.
by nateb2022 2mo ago
Can Sprite replace Pepsi?
56.
by nateb2022 2mo ago
There's some potential research suggesting it has positive long term cardiovascular benefits, including decreased risk of heart attack, stroke, and overall cardiovascular mortality. Anecdotally, a lot of people use it off-label to impr
57.
by nateb2022 2mo ago
OpenRouter already takes in around $140M in yearly revenue. How would paying 5% of yearly revenue make any sense for an acquisition??
58.
Ornith 1.5 35B-A3B: SOTA 35B MoE model(huggingface.co)
3 points by nateb2022 2mo ago|0 comments
59.
OpenAI Partnering with CodeAI to prepare the first AI (human) generation(openai.com)
4 points by nateb2022 2mo ago|0 comments
60.
Amazon Leo plans to connect mobile devices from space(aboutamazon.com)
2 points by nateb2022 2mo ago|0 comments

More ›