Skip to main content
Activity logs provide detailed usage statistics and analytics for your LLM Gateway usage. Query by date range, project, or API key.

​
Get Activity Data

Retrieve aggregated activity statistics grouped by day.
​
days
number
Number of days to retrieve (overridden if from and to are specified)
​
from
string
Start date in YYYY-MM-DD format (inclusive)
​
to
string
End date in YYYY-MM-DD format (inclusive)
​
projectId
string
Filter by specific project ID
​
apiKeyId
string
Filter by specific API key ID
​
activity
array
Array of daily activity objects
curl "https://api.llmgateway.io/activity?days=7" \
  -H "Authorization: Bearer YOUR_SESSION_TOKEN"

{
  "activity": [
    {
      "date": "2024-01-15",
      "requestCount": 1250,
      "inputTokens": 125000,
      "outputTokens": 75000,
      "cachedTokens": 25000,
      "totalTokens": 200000,
      "cost": 4.85,
      "inputCost": 1.25,
      "outputCost": 3.50,
      "requestCost": 0.05,
      "dataStorageCost": 0.05,
      "imageInputCost": 0,
      "imageOutputCost": 0,
      "cachedInputCost": 0,
      "errorCount": 12,
      "errorRate": 0.96,
      "cacheCount": 50,
      "cacheRate": 4.0,
      "discountSavings": 0.50,
      "creditsRequestCount": 500,
      "apiKeysRequestCount": 750,
      "creditsCost": 2.00,
      "apiKeysCost": 2.85,
      "creditsDataStorageCost": 0.02,
      "apiKeysDataStorageCost": 0.03,
      "modelBreakdown": [
        {
          "id": "gpt-4o",
          "provider": "openai",
          "requestCount": 500,
          "inputTokens": 50000,
          "outputTokens": 30000,
          "totalTokens": 80000,
          "cost": 2.00
        },
        {
          "id": "claude-3-5-sonnet-20241022",
          "provider": "anthropic",
          "requestCount": 400,
          "inputTokens": 40000,
          "outputTokens": 25000,
          "totalTokens": 65000,
          "cost": 1.95
        },
        {
          "id": "gpt-4o-mini",
          "provider": "openai",
          "requestCount": 350,
          "inputTokens": 35000,
          "outputTokens": 20000,
          "totalTokens": 55000,
          "cost": 0.90
        }
      ]
    },
    {
      "date": "2024-01-14",
      "requestCount": 1100,
      "inputTokens": 110000,
      "outputTokens": 66000,
      "cachedTokens": 22000,
      "totalTokens": 176000,
      "cost": 4.25,
      "inputCost": 1.10,
      "outputCost": 3.08,
      "requestCost": 0.04,
      "dataStorageCost": 0.03,
      "imageInputCost": 0,
      "imageOutputCost": 0,
      "cachedInputCost": 0,
      "errorCount": 8,
      "errorRate": 0.73,
      "cacheCount": 45,
      "cacheRate": 4.09,
      "discountSavings": 0.45,
      "creditsRequestCount": 440,
      "apiKeysRequestCount": 660,
      "creditsCost": 1.70,
      "apiKeysCost": 2.55,
      "creditsDataStorageCost": 0.01,
      "apiKeysDataStorageCost": 0.02,
      "modelBreakdown": [
        {
          "id": "gpt-4o",
          "provider": "openai",
          "requestCount": 450,
          "inputTokens": 45000,
          "outputTokens": 27000,
          "totalTokens": 72000,
          "cost": 1.80
        },
        {
          "id": "claude-3-5-sonnet-20241022",
          "provider": "anthropic",
          "requestCount": 380,
          "inputTokens": 38000,
          "outputTokens": 23000,
          "totalTokens": 61000,
          "cost": 1.83
        },
        {
          "id": "gpt-4o-mini",
          "provider": "openai",
          "requestCount": 270,
          "inputTokens": 27000,
          "outputTokens": 16000,
          "totalTokens": 43000,
          "cost": 0.62
        }
      ]
    }
  ]
}

​
Understanding Costs

​
Cost Breakdown

Total cost includes:
  1. Input Cost: Tokens sent to the model
  2. Output Cost: Tokens generated by the model
  3. Request Cost: Per-request fees (some providers)
  4. Data Storage Cost: Log retention charges
  5. Image Costs: Vision model image processing
  6. Cached Input Cost: Cached prompt tokens (some providers)

​
Credits vs Provider Keys

The API tracks costs separately based on how requests were processed:
  • Credits Cost: Requests routed through Gateway-managed providers
  • API Keys Cost: Requests using your configured provider keys
This helps you understand:
  • How much you’re spending on credits
  • How much you’re saving by using your own keys
  • Which mode (hybrid, credits, api-keys) is most cost-effective

​
Discounts

Discounts can be applied at the:
  • Global level: All users for a specific provider/model
  • Organization level: Your organization for specific providers/models
The discountSavings field shows how much you saved from applied discounts.

​
Model Breakdown

The modelBreakdown array provides per-model statistics: 
{
  "id": "gpt-4o",
  "provider": "openai",
  "requestCount": 500,
  "inputTokens": 50000,
  "outputTokens": 30000,
  "totalTokens": 80000,
  "cost": 2.00
}
Use this to:
  • Identify your most-used models
  • Compare costs across providers
  • Optimize model selection for cost
  • Track usage patterns

​
Performance Metrics

​
Error Rate

errorRate = (errorCount / requestCount) * 100
A high error rate may indicate:
  • Provider API issues
  • Invalid requests
  • Rate limiting
  • Authentication problems

​
Cache Hit Rate

cacheRate = (cacheCount / requestCount) * 100
A higher cache rate means:
  • Lower costs (cached responses are free)
  • Faster response times
  • Reduced provider API load
Enable caching on your projects to improve cache hit rates and reduce costs.

​
Filtering Examples

​
By Project

Track usage for a specific project: 
curl "https://api.llmgateway.io/activity?projectId=proj_xyz789&days=30" \
  -H "Authorization: Bearer YOUR_SESSION_TOKEN"

​
By API Key

Monitor a specific API key’s usage: 
curl "https://api.llmgateway.io/activity?apiKeyId=key_xyz789&from=2024-01-01&to=2024-01-31" \
  -H "Authorization: Bearer YOUR_SESSION_TOKEN"

​
Custom Date Range

Query activity for a specific time period: 
curl "https://api.llmgateway.io/activity?from=2024-01-01&to=2024-01-15" \
  -H "Authorization: Bearer YOUR_SESSION_TOKEN"

​
Data Retention

Activity data is retained based on your organization’s plan:
PlanRetention Period
Free3 days
Pro90 days
EnterpriseCustom (up to 1 year)
Upgrade to Pro or Enterprise for extended data retention and historical analytics.

​
Error Responses

{
  "message": "Unauthorized"
}

​
Best Practices

Monitor Regularly
  • Check activity daily to catch anomalies
  • Set up alerts for unusual spending
  • Review error rates weekly
Optimize Costs
  • Analyze model breakdown to find expensive models
  • Enable caching for repetitive requests
  • Use cheaper models where appropriate
  • Leverage discounts when available
Track Performance
  • Monitor error rates over time
  • Compare cache hit rates across projects
  • Identify slow or unreliable providers

Build docs developers (and LLMs) love

Get started for freeTalk to us