How to run Qwen 3.8 27B on your laptop for $0 💻
Why now: Simon Willison just had this 17GB open model read raw model weights (a test from an MIT paper). It got 3 of 4 games right. Apache-2.0, 8M downloads in 30 days. 1. Install LM Studio (free) and search "Qwen3.8-27B" 2.
AgenticLinux (Fedora 45 based) now works on DGX Spark 🚀
128 GB of unified memory, a local Qwen 3.8 27B, and OpenClaw doing real work on an immutable box via llmman: steipete jonmasters nullr0ute might be interested in this one.
Reminder that local ai is also important.
This is Qwen 3.8 flash-next. The entire generation ran locally on my 5090 from the video to the actual code. I can’t wait for Qwen 4 series.
If you’re stuck on one llm engine assuming it’s the best, check again
Tested qwen 3.8 27b on 3090 (350w, 64k ctx) with 4 serving engines hyperqwen: fastest at everything except loading. 180 tok/s avg, and 448 tok/s total at 8 users tensorfold: same exl3 files as exllamav3 but
You have reached the end of the archive
All of qwen38