> ## Documentation Index
> Fetch the complete documentation index at: https://docs.memorylake.ai/llms.txt
> Use this file to discover all available pages before exploring further.

# Get Usage

> Quota snapshot and per-model consumption over a period

```
GET /openapi/memorylake/admin/v1/usage
```

Returns the team's quota and consumption. Two different time semantics in one response:

* `quota` is an **account-level snapshot taken now** — what the team can still spend
* `totals` and `by_model` are **aggregates over the requested period**

<Note>
  **Permission required:** `usage:view_all`
</Note>

### Query Parameters

<ParamField query="start_date" type="string">
  First day included, `YYYY-MM-DD`. Defaults to 6 days before `end_date`.
</ParamField>

<ParamField query="end_date" type="string">
  Last day included, `YYYY-MM-DD`. Defaults to today. The window may span at most 92 days.
</ParamField>

<RequestExample>
  ```bash cURL theme={null}
  curl 'https://app.memorylake.ai/openapi/memorylake/admin/v1/usage?start_date=2026-08-01&end_date=2026-08-26' \
    -H 'Authorization: Bearer sk-xxxxxx'
  ```
</RequestExample>

### Response

<ResponseField name="data" type="object">
  <Expandable title="Usage object">
    <ResponseField name="period" type="object">The period the totals cover (`start_date`, `end_date`)</ResponseField>

    <ResponseField name="quota" type="object">
      <Expandable title="Quota snapshot">
        <ResponseField name="available_tokens" type="integer">Currently available tokens. Account-level snapshot, not scoped to the period.</ResponseField>
        <ResponseField name="unlimited" type="boolean">When true, `available_tokens` is not meaningful</ResponseField>
      </Expandable>
    </ResponseField>

    <ResponseField name="totals" type="object">Aggregate `requests`, `prompt_tokens`, `completion_tokens` over the period</ResponseField>

    <ResponseField name="by_model" type="array">
      Per-model breakdown. Entries may name a platform metering operation (`files_process`, `memory_input`, `retrieval_call`, `search_call`) rather than an LLM — those are real billable operations.
    </ResponseField>
  </Expandable>
</ResponseField>

<ResponseExample>
  ```json Success (200) theme={null}
  {
    "success": true,
    "message": "Operation completed successfully",
    "data": {
      "period": { "start_date": "2026-08-01", "end_date": "2026-08-26" },
      "quota": { "available_tokens": 8421000, "unlimited": false },
      "totals": { "requests": 1204, "prompt_tokens": 3210000, "completion_tokens": 1150000 },
      "by_model": [
        { "model": "claude-sonnet-5", "requests": 800, "prompt_tokens": 2400000, "completion_tokens": 900000 },
        { "model": "memory_input", "requests": 404, "prompt_tokens": 810000, "completion_tokens": 250000 }
      ]
    }
  }
  ```
</ResponseExample>