Token Usage
Monitor and control your LLM token spend. Track daily costs, set alerts and hard limits, detect anomalies, and receive automated spend reports.
Overview
The Token Usage page gives you full visibility into token costs and the tools to control them proactively. Rather than reacting to unexpected bills, you can set guardrails that notify you or automatically stop agents before costs exceed your thresholds.
Cost Analytics
Daily cost chart broken down by agent, with monthly projections and trend comparisons.
Spend Alerts
Configurable notifications when spend crosses a threshold. Global, per-environment, or per-agent.
Hard Limits
Automatically pause agent runs when a spending threshold is hit. Other agents continue unaffected.
Anomaly Detection
Automatic alerts when a single agent run costs significantly more than its historical average.
Daily Cost Chart
The chart at the top of the Token Usage page shows a stacked bar for each day of the month. Each bar is broken down by agent, so you can see at a glance which agents are driving costs. Judge evaluation tokens are included in each agent's total.
Use the month selector to navigate between months. The chart goes back to the month the project was created.
Summary Cards
Above the chart, three summary cards provide key metrics at a glance:
| Card | Description |
|---|---|
| Current Total | Total token cost for the selected month. For the current month, also shows the projected total based on daily average so far. |
| Trend | Dollar and percentage difference compared to the previous month. Helps you spot whether costs are growing or shrinking. |
| Active Agents | Number of unique agents that had token usage during the selected month. |
Agent Ranking
Below the chart, a ranking table shows all agents ordered by total spend for the selected month. The top 5 are shown by default, with an option to expand and see all.
Each agent row includes a full token cost breakdown:
| Column | Description |
|---|---|
| Total Cost | Combined cost across all token categories including judge evaluations. |
| Input | Cost of prompt/input tokens (excluding cached portion). |
| Output | Cost of the model's text response tokens. |
| Thinking | Cost of reasoning/thinking tokens (when enabled). |
| Cached | Cost of cached input tokens (billed at reduced provider rate). |
| Judge | Cost of judge evaluation runs associated with this agent. |
Spend Alerts & Limits
Spend alerts and limits let you define thresholds that trigger actions when crossed. Each entry has a type, scope, period, and threshold amount in USD.
Alert vs. Limit
Alert
Notification onlySends a notification (email and/or in-app) when the threshold is crossed. Agents continue running normally. Use alerts for awareness without impacting operations.
Limit
Hard stopSends a notification and pauses new agent runs that fall under this limit's scope. Queued runs are held until the limit is raised, removed, or a new period begins.
Scope
Each alert or limit can be scoped to control exactly what it applies to:
| Scope | Behavior |
|---|---|
| Global | Applies to all token spend across the entire project. When a global limit is hit, all agent runs are paused. |
| Environment | Applies to one or more specific environments. When hit, only runs in those environments are affected. Useful for separating production and staging budgets. |
| Agent | Applies to one or more specific agents by name. When hit, only those agents are paused. All other agents continue normally. |
Period
Choose whether the threshold resets daily or monthly:
- Monthly: Spend is accumulated from the 1st of the month. Resets on the 1st of the next month.
- Daily: Spend resets at midnight UTC each day.
Progress Tracking
Each alert and limit shows a visual progress bar indicating how much of the threshold has been consumed. The bar changes color as spend approaches the limit: default at low usage, amber when nearing the threshold, and red when exceeded.
How Limits Release Runs
When a limit is reached, affected runs are queued with a reference to that specific limit. Runs are automatically released when the limit is deleted, the threshold is raised above current spend, the limit is disabled, or a new period begins (e.g. new month). A global banner appears in the project to inform all team members that a budget limit is active.
Cost Anomaly Detection
When enabled, Connic automatically monitors the cost of each completed agent run and compares it against that agent's 30-day rolling average. If a run costs significantly more than normal, an alert is sent.
Configuration
| Setting | Description |
|---|---|
| Enabled | Toggle anomaly detection on or off for the project. |
| Multiplier | The threshold multiplier. Default is 3x, meaning a run is flagged if it costs more than 3 times the rolling average. Adjust higher to reduce noise or lower for tighter monitoring. |
Minimum Data Requirement
Anomaly detection requires at least 5 completed runs for a given agent in the past 30 days to establish a baseline. Agents with fewer runs are skipped to avoid false positives.
Spend Reports
Automated email reports summarize token costs over a given period. Two report types are available:
Weekly Report
Sent every Monday at 9:00 UTC. Covers the previous 7 days. Includes total cost and a breakdown of the top 5 agents by spend.
Monthly Report
Sent on the 1st of each month at 9:00 UTC. Covers the previous month. Includes total cost and a breakdown of the top 5 agents by spend.
Enable or disable each report type from the Token Usage page. Delivery preferences (which team members receive the email) can be managed in Settings > Notifications, where spend reports appear as dedicated notification types.
Notification Settings
Budget-related events are integrated into the project notification system. Each team member can independently control which budget notifications they receive and through which channel.
| Event Type | In-App | Description | |
|---|---|---|---|
| Budget Alert Triggered | Yes | Yes | A spend alert threshold was crossed. |
| Budget Limit Triggered | Yes | Yes | A hard limit was hit and agents are being paused. |
| Cost Anomaly Detected | Yes | Yes | An agent run cost significantly exceeded its rolling average. |
| Weekly Spend Report | Yes | N/A | Email-only weekly cost summary. |
| Monthly Spend Report | Yes | N/A | Email-only monthly cost summary. |
All notification preferences are managed under Settings > Notifications. Spend reports are email-only and do not appear in the in-app notification feed.
How Costs Are Calculated
Token costs are estimated using the model pricing configured in your project. Connic matches each run's model name against your pricing rules (project-specific exact match, then regex, then global exact, then global regex) and applies the per-million-token rates with volume tiering if configured.
The four token categories (input, output, thinking, cached input) are priced independently. Cached input tokens are a subset of input tokens and are charged at the reduced cached rate instead of the full input rate.
Pricing Configuration
Model pricing is managed under Settings > Model Pricing. Connic ships with global default pricing for popular models. You can override pricing at the project level for custom or fine-tuned models, and define regex patterns to match model name variants.
Check Frequency
Spend alerts and limits are evaluated every 2 minutes. This means there can be a short delay between crossing a threshold and the notification or enforcement taking effect. Anomaly detection runs immediately when each agent run completes.
When you update or delete a limit, blocked runs are released immediately without waiting for the next check cycle.
Full control over your token spend
With daily cost analytics, configurable alerts and limits, anomaly detection, and automated reports, you can stay ahead of unexpected costs and keep your LLM spend predictable.