Token Analytics

Token Analytics provides detailed tracking of AI model usage, costs, and trends across your UniSync instance.

Overview

Every AI model call in UniSync is tracked, including:

Tokens consumed (input + output)
Cost per call
Model used (GPT-4, Gemini, Claude, etc.)
Pipeline step that made the call
Associated agent and environment

Dashboard

The Token Analytics page shows:

Summary Cards

Total Tokens — Lifetime token consumption
Total Cost — Cumulative spending across all providers
Average Cost per Article — Cost efficiency metric
Active Models — Currently used AI models

Usage by Model

Breakdown of token usage per AI model:

Input tokens vs. output tokens
Cost per model
Trend over time

Usage by Pipeline Step

See which phases consume the most tokens:

Content generation (typically the largest)
Source research
Content strategy generation
Keyword classification

Usage by Provider

Aggregate view by provider (OpenAI, Google, Anthropic).

Cost Trends

Time-series charts showing:

Daily/weekly/monthly spending
Cost per article trend
Token consumption trend

Model Pricing

UniSync automatically syncs model pricing data every 24 hours. The model_pricing table maintains current rates for:

Input token price (per 1K tokens)
Output token price (per 1K tokens)
Model capabilities and limits

Filtering

Filter analytics by:

Date range — Custom time periods
Environment — Per-site breakdown
Agent — Per-agent costs
Model — Specific AI model
Provider — OpenAI, Google, Anthropic

API Management — Manage API credentials
AutoPilot — Monitor automated costs
Token Analytics API — API endpoints