AI usage & budget

What the AI has cost, and how much of this month is left in the budget.

The usage summary

As returned by GET /ai/usage. Every endpoint below that returns this resource returns exactly these fields.

Name	Type	Description
`days`required	`integer`	Window length in days: 7, 30 or 90
`total_cost_usd`required	`number`	Billed spend over the window
`total_calls`required	`integer`	Calls over the window
`total_requests`required	`integer`	Provider requests over the window
`total_input_tokens`required	`integer`	Tokens sent over the window
`total_output_tokens`required	`integer`	Tokens returned over the window
`total_cached_input_tokens`required	`integer`	Cached input tokens over the window
`daily`required	`object[]`	Zero-filled, oldest day first
`daily[].date`required	`string`	The day, as YYYY-MM-DD in UTC
`daily[].cost_usd`required	`number`	Billed spend that day
`daily[].calls`required	`integer`	Calls that day
`daily[].requests`required	`integer`	Provider requests that day
`daily[].input_tokens`required	`integer`	Tokens sent that day
`daily[].output_tokens`required	`integer`	Tokens returned that day
`daily[].cached_input_tokens`required	`integer`	Cached input tokens that day
`by_feature`required	`object[]`	Costliest first
`by_feature[].key`required	`string`	The feature, model or source this row aggregates
`by_feature[].cost_usd`required	`number`	Billed spend
`by_feature[].calls`required	`integer`	Assistant ticks and one-shot calls
`by_feature[].requests`required	`integer`	Provider HTTP requests
`by_feature[].input_tokens`required	`integer`	Tokens sent
`by_feature[].output_tokens`required	`integer`	Tokens returned
`by_feature[].cached_input_tokens`required	`integer`	Input tokens served from the provider cache
`by_model`required	`object[]`	Costliest first
`by_model[].key`required	`string`	The feature, model or source this row aggregates
`by_model[].cost_usd`required	`number`	Billed spend
`by_model[].calls`required	`integer`	Assistant ticks and one-shot calls
`by_model[].requests`required	`integer`	Provider HTTP requests
`by_model[].input_tokens`required	`integer`	Tokens sent
`by_model[].output_tokens`required	`integer`	Tokens returned
`by_model[].cached_input_tokens`required	`integer`	Input tokens served from the provider cache
`by_source`required	`object[]`	business is your own provider key; platform is capacity we served
`by_source[].key`required	`string`	The feature, model or source this row aggregates
`by_source[].cost_usd`required	`number`	Billed spend
`by_source[].calls`required	`integer`	Assistant ticks and one-shot calls
`by_source[].requests`required	`integer`	Provider HTTP requests
`by_source[].input_tokens`required	`integer`	Tokens sent
`by_source[].output_tokens`required	`integer`	Tokens returned
`by_source[].cached_input_tokens`required	`integer`	Input tokens served from the provider cache

Endpoints

2 endpoints, each on its own page with its parameters, responses, error codes and samples in curl, Node and Python.

The usage summarylink

Endpointslink

The usage summary

Endpoints