Using completion()/cost_per_tokenI) with LiteLLM Proxy to Return Cost from litellm_config.yaml

[sujeet-turinton]

I’ve set up a LiteLLM proxy with litellm_config.yaml where I defined my custom model (ollama/llama3) along with its prompt_cost_per_million and completion_cost_per_million.
When I call the proxy directly with curl, I get usage tokens back, but the Python client call: always shows None for response_cost.
My question is:
Can the completion() function automatically use the proxy configs (pricing set in litellm_config.yaml) so that cost shows up in response._hidden_params["response_cost"]?
Or do I need to explicitly call cost_per_token() in my script and re-define pricing there?
Basically, I want my client-side code to reflect the cost defined in the proxy config.
Thanks a lot!
Would appreciate help!.