Performance Benchmark

Measure the prompt-compressor /compress endpoint directly. Edge routing is not included.

-Progress
-Browser p50
-LLMLingua p50
- Model requests - LLMLingua chunks
-Errors
Enter an API key with Use Compression permission or start a bounded demo session. Credentials stay in page memory and out of downloads.
Ready

Summary

No results
Case Runs Browser p50 Server p50 Preprocess p50 Selection p50 Model requests Chunks Gate skips Top gate LLMLingua p50 Token est. p50 Diagnostics p50 Reduction avg

Runs

0 rows
Case Repeat Status Browser Server Preprocess LLMLingua In tokens Out tokens Mode Path Gate Error

Log

Idle