Qwen 3.8 blows mind - and this time it's real

Article automatically generated from technical news.

Edit: for anyone interested - that's a q8 Qwen 3.8 27b running on double v100 totaling 64gb, 256k context window, q8 kv quantization. I never saw a behavior like that from any other local LLM - can't vouch for global since we never see what's going on under the hood. Anyways - Qwen 3.8 got a task to go through a lot of files (over 700) and sort them out by content. It never spun subagents. It wrote and ran a python script to prompt itself. I believe thats the most brilliant behavior