k__

vor 7 Stundendiscuss

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__

vorgesterndiscuss

[dead]

k__

vor 6 Tagendiscuss

"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__

vor 7 Tagendiscuss

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__

vor 9 Tagendiscuss

Thanks!

I'll try that.

k__

vor 9 Tagendiscuss

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__

vor 9 Tagendiscuss

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__

vor 11 Tagendiscuss

Llama3.1 instruct seems to be doing okay on that page, mostly because it's dirt cheap.

k__

vor 12 Tagendiscuss

So, their incentive is to promote AI music, since they don't have to pay royalties for them.

k__

vor 23 Tagendiscuss

Did Nim 3 go all-in on compile time memory management?

k__

vor 23 Tagendiscuss

I'm curious about Denuvo's opinion on that.

k__

vor 23 Tagendiscuss

Maybe, you could pipe it through T5 or something.

k__

vor 24 Tagendiscuss

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__

vor 24 Tagendiscuss

As I understood it, they were only a US thing anyway. Because of healthcare issues etc.

k__

vor 24 Tagendiscuss

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__

vor 25 Tagendiscuss

I thought the joke was that people aren't paying enough money.

k__

vor 27 Tagendiscuss

100K seems quite much.

I had the impression, models would get inconsistent after just 3000 words.

k__

letzten Monatdiscuss

Why doesn't Mistral distill?

k__

letzten Monatdiscuss

I tried it with embedded programming, and failed miserably.