nomyo-ai/nomyo-router
3
2
Fork 0
Code Issues 1 Pull requests 8 Projects Releases 10 Packages 1 Wiki Activity Actions

token usage counter for non-stream openai ollama endpoints added

Browse source
This commit is contained in:
Alpha Nerd 2025-11-06 14:27:34 +01:00
parent f0f6069577
commit 47a39184ad
1 changed files with 6 additions and 0 deletions
Download patch file Download diff file Expand all files Collapse all files

6
router.py
View file

@ -1446,6 +1446,9 @@ async def openai_chat_completions_proxy(request: Request):
yield f"data: {data}\n\n".encode("utf-8")
yield b"data: [DONE]\n\n"
else:
prompt_tok = async_gen.usage.prompt_tokens or 0
comp_tok = async_gen.usage.completion_tokens or 0
record_token_usage(endpoint, payload.get("model"), prompt_tok, comp_tok)
json_line = (
async_gen.model_dump_json()
if hasattr(async_gen, "model_dump_json")
@ -1549,6 +1552,9 @@ async def openai_completions_proxy(request: Request):
# Final DONE event
yield b"data: [DONE]\n\n"
else:
prompt_tok = async_gen.usage.prompt_tokens or 0
comp_tok = async_gen.usage.completion_tokens or 0
record_token_usage(endpoint, payload.get("model"), prompt_tok, comp_tok)
json_line = (
async_gen.model_dump_json()
if hasattr(async_gen, "model_dump_json")