Qwen 3.8 max just ran an AI project for 10 days without a human.
And that’s not even the most impressive test Alibaba showed. What it actually did: → Started with an empty folder and built a working CLI tool from scratch → Created its own job queue, ran tests, caught
On the flight I tried MTPLX and ds4 with Qwen 3.8 Flash Next on M5 Max in Low Power.
Not bad! In ds4 I've not cooked support for MTP in ds4 chat. - MTPLX 25 t/s - DS4 18 t/s
Qwen 3.8: 27b on fire 🔥 someone is burning 1500tok/sec.
What are people getting w/ Qwen-3.8-27B on M5 Pro with 24gb?
I think we might be close to 2x'ing (this: with the iPhone handling spec decode and mac handling verification. probably some room to improve here. Just glad that my 24gb machine might not be
You have reached the end of the archive
All of qwen38