I'll be testing Alibaba’s new Qwen 3.8 Flash Next model, which is an experimental preview of the upcoming Qwen 4 architecture.
It’s a 125-billion-parameter MoE model that only activates 6 billion parameters per token, making it extremely cheap and efficient, while still offering
You can now run a legit 27B parameter model completely offline on a normal consumer PC🚀
Qwen 3.8 27B. Download LM Studio (free) → search “Qwen 3.8” → official version from the Qwen team on Hugging Face. They have 4-bit, 5-bit, 6-bit and 8-bit quantizations. Higher bits =
Used Qwen 3.8 27b for it all (local) with OMP.
I love making silly things like this, and AI allows me to move so much quicker (and with better art). Link below.
Qwen 3.8 27B did a better job than 3.8 Flash Next, but it is still not even close.
I need to try Grok and Codex to see if this is just a really hard game to recreate. I am trying 3.8 Flash Next one last time using DeepSeek Harness because I've had really good luck with it.
You have reached the end of the archive
All of qwen38