Skip to content
vitest-evals

UsageSummary

UsageSummary = object

Provider usage summary attached to a normalized harness run.

Example

Section titled “Example”
const usage: UsageSummary = {
  provider: "openai",
  model: "gpt-4o-mini",
  inputTokens: 212,
  outputTokens: 48,
  totalTokens: 260,
};

Properties

Section titled “Properties”

inputTokens?

Section titled “inputTokens?”

optional inputTokens?: number

Input, prompt, or request tokens consumed by the run.

metadata?

Section titled “metadata?”

optional metadata?: Record<string, JsonValue>

Provider-specific JSON-safe usage details. Cost estimates belong here.

model?

Section titled “model?”

optional model?: string

Model used for the application run.

outputTokens?

Section titled “outputTokens?”

optional outputTokens?: number

Output or completion tokens produced by the run.

provider?

Section titled “provider?”

optional provider?: string

Provider that served the application run.

reasoningTokens?

Section titled “reasoningTokens?”

optional reasoningTokens?: number

Reasoning tokens reported by providers that expose them.

retries?

Section titled “retries?”

optional retries?: number

Retry count observed during the run.

toolCalls?

Section titled “toolCalls?”

optional toolCalls?: number

Count of tool calls observed during the run.

totalTokens?

Section titled “totalTokens?”

optional totalTokens?: number

Total token count reported by the provider or adapter.