Skip to content

Compress tool outputs, logs, files, and RAG chunks before they reach the LLM. 20% fewer tokens for coding agents, 60-95% fewer tokens for JSON, same answers. Library, proxy, MCP server.

Fix inflated token savings stats from cache hits

chopratejas1aa312b92300
diff --git a/headroom/proxy/server.py b/headroom/proxy/server.py
index 72e9e4c75..03392b44d 100644
--- a/headroom/proxy/server.py
+++ b/headroom/proxy/server.py
@@ -1546,7 +1546,7 @@ class HeadroomProxy:
model=model,
input_tokens=0,
output_tokens=0,
- tokens_saved=cached.tokens_saved_per_hit,
+ tokens_saved=0, # Savings already counted when response was cached
latency_ms=optimization_latency,
cached=True,
)
@@ -3957,7 +3957,7 @@ class HeadroomProxy:
model=model,
input_tokens=0,
output_tokens=0,
- tokens_saved=cached.tokens_saved_per_hit,
+ tokens_saved=0, # Savings already counted when response was cached
latency_ms=(time.time() - start_time) * 1000,
cached=True,
)