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(backends/litellm): guard None completion_tokens in usage mapping (#2322)

Abhay Singh44a174fef4d5
diff --git a/headroom/backends/litellm.py b/headroom/backends/litellm.py
index 36987b3e5..08e972908 100644
--- a/headroom/backends/litellm.py
+++ b/headroom/backends/litellm.py
@@ -445,7 +445,12 @@ def _anthropic_usage_from_litellm(litellm_usage: Any) -> dict[str, Any]:
prompt_tokens = int(getattr(litellm_usage, "prompt_tokens", 0) or 0)
usage: dict[str, Any] = {
"input_tokens": max(prompt_tokens - cache_read - cache_write, 0),
- "output_tokens": getattr(litellm_usage, "completion_tokens", 0),
+ # None-guard like the other fields: LiteLLM's Usage always carries the
+ # completion_tokens attribute, so the getattr default never fires, but a
+ # provider can leave it None. Emitting output_tokens=None would break the
+ # RequestOutcome int contract downstream (e.g. prometheus does
+ # tokens_output_total += output_tokens -> TypeError).
+ "output_tokens": int(getattr(litellm_usage, "completion_tokens", 0) or 0),
}
if cache_read or cache_write:
usage["cache_read_input_tokens"] = cache_read
diff --git a/tests/test_litellm_nonstream_cache_usage.py b/tests/test_litellm_nonstream_cache_usage.py
index 872456bcf..1943669b6 100644
--- a/tests/test_litellm_nonstream_cache_usage.py
+++ b/tests/test_litellm_nonstream_cache_usage.py
@@ -72,3 +72,15 @@ def test_input_tokens_never_negative() -> None:
)
)
assert usage["input_tokens"] == 0
+
+
+def test_output_tokens_none_coerced_to_zero() -> None:
+ # A provider can carry the completion_tokens attribute but leave it None.
+ # The mapping must emit an int (0), not None, so RequestOutcome's int
+ # contract holds downstream (prometheus does tokens_output_total +=
+ # output_tokens, which would raise TypeError on None).
+ usage = _anthropic_usage_from_litellm(
+ SimpleNamespace(prompt_tokens=100, completion_tokens=None)
+ )
+ assert usage["output_tokens"] == 0
+ assert isinstance(usage["output_tokens"], int)