Dashboard

Using the dashboard

The LLM Cost Tracker dashboard gives you seven views into your LLM spend. All views filter to the currently selected project.

Overview

The main dashboard shows four KPI cards at the top:

• Total spend — total USD spent on LLM calls for this project
• Total calls — number of tracked LLM calls
• Avg cost / call — average cost per call. Watch this number after deploys — a prompt change that adds tokens will show up here immediately.
• Avg tokens / call — average total tokens per call

Below the KPIs you'll find a spend-by-group breakdown with a model composition panel, followed by the full call log.

By group

Groups all calls by their group tag and shows total cost, call count, average tokens, average cost per call, and average latency per group. The percentage column shows what share of your total bill each group represents.

This is the fastest way to answer: "which part of my app is driving my LLM bill?"

By user

Groups all calls by userId and ranks users by total spend. Useful for:

• Validating your pricing model — are your highest-cost users on the right plan?
• Detecting abuse — unusual call volumes from a single user
• Unit economics — cost to serve per user vs revenue per user

Spend targets

Set a monthly dollar target per group and track actual vs plan. The Spend targets page shows:

• Targeted groups — groups with a target set, showing MTD spend, target, projected month-end spend, and status (On track, Watch, or Over target).
• Unplanned spend — groups generating spend with no target set. Add a target to move them above.

Targets are per calendar month. Projections assume current daily burn rate holds through end of month. See Spend targets for full details.

Enforcement

Manage spend limits and view enforcement history. The Enforcement page has three panels:

• Tier templates — spend limits that apply automatically to all users on a given tier. Configure once, enforced for everyone who matches.
• Per-user overrides — limits for specific user IDs that take precedence over any matching tier template.
• Enforcement log — a feed of every blocked, warned, or dry-run event with timestamp, user, spend at time, and limit.

Budget enforcement requires a Growth plan. See Budget enforcement for full details.

Call log

A real-time feed of every LLM call, most recent first. Shows timestamp, group, model, user, input tokens, output tokens, cost, and latency. Limited to the last 100 events — use the By group and By user views for aggregate analysis.

Versions

Compares cost per call across prompt versions. Tag your deploys with a promptVersion string — a git SHA or label like "v2.1" — and this view shows you exactly how each prompt change affected cost per call, token usage, and latency side by side.

Settings

Manage your projects and view API keys. Each project has its own API key — use separate projects for separate apps or environments (e.g. myapp-prod and myapp-staging).

Next: By group →