compass.utilities.costs.cost_for_model#

cost_for_model(model_name, prompt_tokens, completion_tokens)[source]#

Compute the API costs for a model given the token usage

Parameters:

  • model_name (str) – Name of the model. Needs to be registered as a key in LLM_COST_REGISTRY for this method to return a non-zero value.

  • prompt_tokens (int) – Number of prompt and completion tokens used, respectively.

  • completion_tokens (int) – Number of prompt and completion tokens used, respectively.

Returns:

float – Total cost based on the token usage.

On this page