Cut Qwen3.8-27B Reasoning Tokens by 40% -- 3.8 'ThinkingCap' benchmarked!

EDIT: Sorry for the unclear title. This model is UkisAI's Swift-Qwen3.8-27B, not a new version of BottleCap AI's 3.6-ThinkingCap. All credit goes to UkisAI for making great fine-tune, and I made this post to celebrate their work. I meant no disrespect by mentioning another model…

Published 15 Sept 2026, 18:36 · Updated 15 Sept 2026, 23:17