Qwen3.8 flash next + exllamav3 + hermes is amazing
retained text
I know there is nothing new with what I am saying but I recently started with hermes agent (it’s been a while I wanted to but did not have the time). Qwen3.8fn 6bpw exl3 (from turboderp) on a 6x3090 (I assume lower quants on lower number of gpus work same) gives me around 80-120t/s with good pp, and with good quality. That engine is crazy for cuda dude! So now I control hermes from my phone securely (via Matrix) everyday and discover more and more its potential besides delegating to a coding agent/harness (opencode). Qwen + Turboderp + Nous -> love on you
document fetched 2026-09-27 00:39 · canonical https://reddit.com/r/LocalLLaMA/comments/1wqn378/qwen38_flash_next_exllamav3_hermes_is_amazing