mydreamof

3 ms·

1.
Qwen3.8 Max – Cost per task higher than Astra on Artificial Analysis(artificialanalysis.ai)
1 points by mydreamof 2mo ago|1 comments
2.
by mydreamof 2mo ago
Slower, less reliable but cost more :D Of course this is max reasoning so I guess it make no sense to use it
3.
by mydreamof 2mo ago
Seems like benchmaxing? For example for Terminal-Bench 4 it doesn't have great results. And why not show other benchmarks?
4.
by mydreamof 2mo ago
It is more token efficient I believe
5.
Every Chess Position(wojciechspace.com)
2 points by mydreamof 2mo ago|1 comments
6.
by mydreamof 2mo ago
Inspired by everyuuid.com What I found interesting is that when you click a random line multiple times there is almost always plenty of queens. On averege there is 5.5 queens because the whole space is dominated by near full boards. If you
7.
by mydreamof 2mo ago
Ofcouse it will go down. Look at cost of Fable vs GPT-6 - it is already 2.3 cheaper
8.
by mydreamof 2mo ago
I don't get it. For me it seems Opus was more accurate in terms of for example this small building in the right down corner
9.
by mydreamof 2mo ago
Wow its better than Fable5 and GPT5.6 SOL in many benchmarks
10.
Qwen3.8-Max just got upgraded(twitter.com)
7 points by mydreamof 2mo ago|2 comments
11.
by mydreamof 2mo ago
Big jump in benchmarks
12.
by mydreamof 3mo ago
IMO it depends on the company. If everyone in the company has to provide a lot of PRS etc. you have to stop care about the quality and just get the paycheck. The CEO decided not you. (Of course not when the product may charm ppl etc.)
13.
by mydreamof 4mo ago
Bro these colors on chars are unbelievlable, I can not understand which is opus, which is fable, which is GPT...
14.
It Still Can't Do My Job: Four Years of Moving Goalposts (2022–2026)(publicznyprofil.github.io)
55 points by mydreamof 4mo ago|138 comments
15.
by mydreamof 5mo ago
you can even use gpt 1.0 but it proves nothing
16.
by mydreamof 5mo ago
there is no GPT 5.6 init, so what's the point?
17.
by mydreamof 5mo ago
It would be great to see some benchmark how it improve the agent. Like agent will be better at getting information from such data or what? Otherwise what is the goal of using it