Bottlecap AI releases ThinkingCap for Gemma-4-26B-A4B-it with lower thinking-token use

Bottlecap AI published ThinkingCap-gemma-4-26B-A4B-it, the fifth model in its ThinkingCap series and a token-efficient adaptation of Google’s gemma-4-26B-A4B-it. The company says the model cuts thinking tokens by 38.9% across twelve benchmarks while changing…