HackerLangs
TopNewTrendsCommentsPastAskShowJobs

k__

15,768 karmajoined hace 13 años
Author that tries to smash the gates that keep you from building great things.

Website: https://kay.is

Blog: https://fllstck.dev

GitHub: https://github.com/kay-is

Submissions

Borrow Checking in Nim 3.0

github.com
9 points·by k__·hace 8 meses·0 comments

Nepal turns to Bitcoin and stablecoins as protests rock the nation

thestreet.com
4 points·by k__·hace 10 meses·0 comments

Physicists devise an idea for lasers that shoot beams of neutrinos

news.mit.edu
2 points·by k__·hace 10 meses·0 comments

comments

k__
·hace 8 horas·discuss
Pi
k__
·ayer·discuss
Yeah, DeepSeek V4 (Flash and Pro) is below $0.004 for 1M cache hits.

Even with usage based billing I'm below $1 writing code all day.
k__
·anteayer·discuss
[dead]
k__
·hace 6 días·discuss
"Europe's rapid warming is partly the result of [...] a drop in the number of tiny polluting particles in the air. This means that less of the Sun's energy is reflected back into space, leaving more energy to heat the Earth's surface."

Hmmm...
k__
·hace 7 días·discuss
I tried DS4 Pro yesterday for one particular task that Flash struggled with.

That one Pro task cost $1.50 and the whole day of using Flash cost me $5.
k__
·hace 9 días·discuss
Thanks!

I'll try that.
k__
·hace 9 días·discuss
Yeah, v4 flash is dirt cheap, but it's running in circles quite often.

Might very well be that a better model is cheaper if it gets things right the first try.

Maybe I should route to a better model when v4flash hasn't solved after a specific number of tokens.
k__
·hace 9 días·discuss
Nice.

I paid $6 yesterday for DeepSeek V4 Flash on OpenRouter. That's like $120 dollar for a month, and it's not even a good model.
k__
·hace 11 días·discuss
Llama3.1 instruct seems to be doing okay on that page, mostly because it's dirt cheap.
k__
·hace 12 días·discuss
So, their incentive is to promote AI music, since they don't have to pay royalties for them.
k__
·hace 23 días·discuss
Did Nim 3 go all-in on compile time memory management?
k__
·hace 23 días·discuss
I'm curious about Denuvo's opinion on that.
k__
·hace 23 días·discuss
Maybe, you could pipe it through T5 or something.
k__
·hace 24 días·discuss
Training DeepSeek was magnitudes cheaper than training the SOTA models it relied upon.

In theory, other countries should be able to replicate that effort and improve it.
k__
·hace 24 días·discuss
As I understood it, they were only a US thing anyway. Because of healthcare issues etc.
k__
·hace 25 días·discuss
I tried some smaller Gemma4 and Qwen3.6 quants on my MBA with M5/16GB and had like 20-60 tokens per second. At 60 it felt pretty okay and that hardware is on the lower end.

I'd assume a Mac with 32-64GB memory would get some reasonable results.
k__
·hace 25 días·discuss
I thought the joke was that people aren't paying enough money.
k__
·hace 27 días·discuss
100K seems quite much.

I had the impression, models would get inconsistent after just 3000 words.
k__
·el mes pasado·discuss
Why doesn't Mistral distill?
k__
·el mes pasado·discuss
I tried it with embedded programming, and failed miserably.