Tokensaver, see which parts of your LLM prompt cost money (github.com)

Tokensaver has launched a new tool aimed at optimizing prompts for various AI models by revealing the cost in tokens and dollars associated with each component of the input. Unlike typical token counters, Tokensaver provides actionable suggestions on how to reduce costs, explaining the rationale behind each optimization while ensuring that no data leaves the user’s device, thereby enhancing privacy. This tool is capable of working offline and supports prompts in more than 55 languages, accommodating different scripts, including those without spaces between words. The significance of Tokensaver for the AI/ML community lies in its potential to streamline the use of large language models (LLMs) by significantly lowering operational costs. By optimizing prompts—in areas such as JSON formatting, whitespace management, and removal of redundant phrases—users can achieve up to 20% token savings per call, translating to substantial cost reductions over time, especially for high-volume applications. This approach empowers developers and researchers to refine their interactions with LLMs, facilitating more efficient experimentation and deployment of AI solutions while maintaining control over their data.