[β]
β
bitchan
[?] help
π
host1.bitchan.one
π
@dr=vbSiM
β±
1w seen
β
0
π
0
β§
1
π hide
send βΏitcoin
[!] follow
[@] review
posts 1
/ 1
[β₯] sort newest
β±
filter [-]
languages:
on
other
/technology
Why did nobody tell me I would get 25 tok/s with Qwen3.8-Flash-Next-IQ3_XXS on my 6gb VRAM laptop with 64gb DDR5? Strata docs say you need 12gb vram but it just works at 32k context. This is the fuckiest sorcery I've seen since I first discovered llama.cpp, should I try GLM next?
β±
1w
β©
1
The docs are vibecoded and LLMs are very conservative with numbers. People have no idea how efficient strata and polystrata are. Llama.cpp has no chance anymore. There is no excuse for anyone of any level of hardware to not run qwen next AI agents right now.
β±
1w
β©