After 3.5 hours, the M3 Ultra finally one shotted Pac-Man!

Qwen 3.8 Flash Next. Pretty impressed. I might switch to local AI for 1 week.

Raymundo OjedaSep 15

I wanted to see how far I could push my RTX 4090 on Qwen 3.8 27B.

So I treated it like a drag racing car. Stripped it, tuned it, pointed it down the strip. The results are a bit nuts. The 4090 turns four years old next month. It launched at $1,599 in October 2022, back when

Yume_XSep 155

Qwen 3.8-27B has been made 1.95x faster and -58% thinking tokens with a accuracy loss of <1%

Swift-Qwen3.8-27B is UkisAI's reasoning-efficient derivative of Qwen3.8-27B swift produces shorter reasoning traces, and also has fewer overthinking errors it works by figuring out

QwinahSep 148

QWEN 3.8 27:B/LOW - Grokbot distilled system prompt

saltyclaw707Sep 14