Low-latency prompt compression

Save on LLM input tokens before every request.

Meet frontier models with frontier query planning. Vrbose shortens prompts before they reach your model, helping teams reduce input token costs with a simple API and evolving, cutting-edge compression algorithm.

Create a free account Read the API docs

Free plan: 20/day
Paid plan: $0.01/query
Potential savings: Up to 50%; How cost optimization works →

Say what you want, with less.

Send Vrbose your original query, receive a compressed version, and pass that on to your LLM provider.

Request
  POST /detokenate
  Authorization: Bearer vb_live_...

  { "q": "Please review the following 
  customer support conversation and provide 
  a concise summary. Customer says: I 
  cannot log in." }

Response
  { "q": "summarize support conversation.
  Customer says: I cannot log in." }

Example above provides 48% savings.

Demo of prompt compression with the Vrbose API

So easy, a kid can do it! Takes less than 30 seconds to start saving tokens.