Usage & Cost Analytics

Overview

Ingate tracks every proxied request and computes dollar cost from token counts using a built-in price table. Usage data can be queried as summaries, time-series, or breakdowns across multiple dimensions. All analytics endpoints are scoped to the current organization and require a valid API key.

Analytics data powers both the API responses documented below and the built-in Dashboard Views available in the Ingate console.

Usage Summary

curl "https://api.ingateai.com/api/v1/usage/summary?since=2026-03-01T00:00:00Z" \
  -H "Authorization: Bearer <token>"

{
  "total_requests": 142857,
  "total_input_tokens": 82400000,
  "total_output_tokens": 31200000,
  "total_cost_usd": 1284.62,
  "avg_latency_ms": 842,
  "p99_latency_ms": 3200,
  "cache_hit_rate": 0.23,
  "error_rate": 0.004,
  "since": "2026-03-01T00:00:00Z",
  "until": "2026-04-01T00:00:00Z"
}

Time-Series

Returns usage metrics grouped into fixed-width time buckets. The granularity parameter controls bucket width: hour (default) or day. All summary filters (since, until, app_id,team_id, provider, model, user_id,session_id) apply.

curl "https://api.ingateai.com/api/v1/usage/timeseries?granularity=day&since=2026-03-01T00:00:00Z" \
  -H "Authorization: Bearer <token>"

{
  "granularity": "day",
  "buckets": [
    {
      "timestamp": "2026-03-01T00:00:00Z",
      "requests": 4820,
      "input_tokens": 2800000,
      "output_tokens": 1050000,
      "cost_usd": 42.18,
      "avg_latency_ms": 810,
      "error_count": 12
    },
    {
      "timestamp": "2026-03-02T00:00:00Z",
      "requests": 5104,
      "input_tokens": 3100000,
      "output_tokens": 1180000,
      "cost_usd": 47.93,
      "avg_latency_ms": 795,
      "error_count": 8
    }
  ]
}

Breakdown

The group_by parameter is required. Valid values:

All summary filters apply. Results are sorted by total_cost_usd descending and limited to the top 100 groups by default (use limit to adjust).

# Breakdown by model for the last 7 days
curl "https://api.ingateai.com/api/v1/usage/breakdown?group_by=model&since=2026-03-28T00:00:00Z" \
  -H "Authorization: Bearer <token>"

{
  "group_by": "model",
  "groups": [
    {
      "key": "gpt-4o",
      "requests": 52400,
      "input_tokens": 31000000,
      "output_tokens": 12800000,
      "cost_usd": 420.00
    },
    {
      "key": "claude-sonnet-4-20250514",
      "requests": 28100,
      "input_tokens": 18200000,
      "output_tokens": 7400000,
      "cost_usd": 234.60
    }
  ]
}

# Per-user breakdown for a specific app
curl "https://api.ingateai.com/api/v1/usage/breakdown?group_by=user_id&app_id=app-uuid-here" \
  -H "Authorization: Bearer <token>"

Cost Tracking

Ingate calculates dollar cost for every request using a built-in price table with per-million-token rates for input and output tokens. Costs are computed automatically with no configuration needed. When a request uses a model not in the price table, tokens are still recorded but cost is reported as null.

Supported Models

Price Table API

curl "https://api.ingateai.com/api/v1/cost/prices" \
  -H "Authorization: Bearer <token>"

{
  "updated_at": "2026-04-01T00:00:00Z",
  "prices": [
    {
      "provider": "openai",
      "model": "gpt-4o",
      "input_per_million": 2.50,
      "output_per_million": 10.00
    },
    {
      "provider": "openai",
      "model": "o4-mini",
      "input_per_million": 1.10,
      "output_per_million": 4.40
    },
    {
      "provider": "anthropic",
      "model": "claude-sonnet-4-20250514",
      "input_per_million": 3.00,
      "output_per_million": 15.00
    },
    {
      "provider": "ollama",
      "model": "*",
      "input_per_million": 0,
      "output_per_million": 0
    }
  ]
}

Dashboard Views

The Ingate console includes a built-in analytics dashboard that visualizes usage and cost data in real time. The dashboard is available at https://console.ingateai.com/analytics and requires no additional setup.

Overview Widgets

The top of the dashboard displays summary cards for the selected time range:

• Total Requests: aggregate request count with trend indicator vs. prior period
• Total Tokens: combined input + output token count
• Total Cost: dollar spend (Enterprise only)
• Avg Latency: mean response time in milliseconds
• Cache Hit Rate: percentage of requests served from cache
• Error Rate: percentage of requests returning 4xx/5xx

Each widget shows the current value and a comparison delta against the previous equivalent period (e.g., this week vs. last week).

Charts

Below the widgets, interactive charts provide deeper visibility:

• Requests over time: stacked area chart with hourly or daily granularity, segmented by provider or model
• Token usage over time: line chart showing input vs. output tokens
• Cost over time: bar chart of daily spend, color-coded by provider (Enterprise only)
• Latency distribution: histogram of response times with p50, p95, and p99 markers
• Top models: horizontal bar chart ranking models by request volume or cost
• Breakdown table: sortable table matching the Breakdown API, groupable by app, team, model, provider, user, or session

All charts respect the global time range selector and any active filters (app, team, provider, model). Data refreshes automatically every 60 seconds while the dashboard is open.