BottleCap AI releases two ThinkingCap models with claimed token savings
BottleCap AI has made ThinkingCap-Gemma-4–26B-A4B and ThinkingCap-Qwen3.8-Flash-Next available for free on Hugging Face, including in GGUF format. According to the company, both use approximately 39 % fewer tokens on average, with small changes in accuracy.
BottleCap AI has released ThinkingCap-Gemma-4–26B-A4B and ThinkingCap-Qwen3.8-Flash-Next. The first is based on Gemma 4 from Google, and the second on Qwen 3.8 Flash Next from Alibaba. Both are available for free on Hugging Face, including GGUF variants.
For ThinkingCap-Gemma-4–26B-A4B, the company reports a reduction in token usage of more than 69 % in some cases and nearly 39 % on average. For ThinkingCap-Qwen3.8-Flash-Next, the claimed savings reach up to 51 %, averaging just under 39 %. According to the company, these modifications are accompanied by only small changes in accuracy.
The company also offers paid Pro variants. For the variant derived from Gemma 4, it claims savings of up to 88 % and 62 % on average. For the variant derived from Qwen 3.8 Flash Next, it reports average savings of 60 %.
Why it matters
The difference between maximum and average savings shows that the benefit depends on the task. Decisions about deployment therefore need to be informed by testing both token usage and accuracy on specific tasks; the published percentages are claims made by the company.
Two audiences, two different impacts
What this means
For individuals
Individuals can try the optimized models without purchasing a paid Pro variant and assess whether their accuracy meets their needs.
For a business
Businesses now have the option to compare free and paid variants when selecting a model for production use. Claimed token savings alone are not enough to determine total costs.
DevelopmentCheck the original
Event sources
only one source so far · 1 publisher, 1 independent. We count feeds from the same owner only once.