Launching today
SuperCompress
Cut your LLM Token Costs by 65%
25 followers
Cut your LLM Token Costs by 65%
25 followers
SuperCompress is open-source prompt compression that cuts LLM token costs by ~65%. Reduce input tokens before OpenAI, Claude, Gemini, RAG, chatbot, and agent calls while preserving answer-critical evidence. It's easily integrable into your app via the Supercompress API, so you can save money but keep users happy. It preserves more than 98% of critical content, while removing more than 65% of tokens on average.
Plugged it into a small RAG pipeline and watched the token bill drop almost overnight without the answers getting dumber, which is honestly the part I was most skeptical about.