BottleCap AI Releases ThinkingCap-Qwen3.8-27B: 37.2% Fewer Thinking Tokens at a 0.86pp Accuracy Cost

BottleCap AI launched ThinkingCap-Qwen3.8-27B, a fine-tuned version of Qwen3.8-27B that reduces reasoning tokens by 37.2% with only a 0.86 percentage point drop in accuracy across 12 benchmarks. The model maintains performance on reasoning, instruction following, and safety, and is deployable on multiple platforms with licensing restrictions.