Show HN: Butter – A Behavior Cache for LLMs
6 months ago
- #cost-saving
- #LLM
- #cache
- Butter is a cache for LLM responses that identifies patterns to save costs.
- It's deterministic, ensuring consistent repeatability of past AI behaviors.
- Compatible with Chat Completions API, easily integrates with tools like LangChain, Mastra, and others.
- Targeted at autonomous agents performing repeat tasks like data entry and research.
- Priced at 5% of the savings on token bills, currently free to use.