How to Track AI API Usage and Costs Across Multiple Providers
How to Track AI API Usage and Costs Across Multiple Providers
If you use multiple AI APIs — Claude for coding, Copilot for completions, Mistral for experiments — keeping track of spending and quota usage across providers is a pain. Each has its own dashboard, its own billing page, its own rate limits.
AI Usage Tracker puts all of that into a single compact window.
The problem
Most developers using AI APIs face these issues:
- Scattered dashboards — each provider has its own usage page
- No unified cost view — you have to manually sum spending across providers
- Quota surprises — hitting rate limits mid-work because you didn't check remaining capacity
- No pace projection — knowing you've used 60% of your quota doesn't tell you if you'll run out before the reset window
What AI Usage Tracker does
It's a Windows tray utility (also runs on Linux) that monitors all your AI APIs from one place:
| Feature | Details |
|---|---|
| Unified dashboard | Current spend and token usage across all providers at a glance |
| Pace projection | Estimates whether you'll exhaust your quota before the reset window |
| Tray integration | Auto-hide when focus is lost, individual tray icons per provider |
| Live progress bars | "Core Temp" style bars in the tray icon |
| Auto-discovery | Scans environment variables and Claude Code credentials for API keys |
| Auto-updates | Notifies you of new releases |
Supported providers
| Provider | Integration | Status |
|---|---|---|
| GitHub Copilot | OAuth Device Flow | Tested |
| Anthropic Claude Code | Auto-detected | Partially tested |
| Mistral | API Key | Tested |
| DeepSeek | API Key | Beta |
| Kimi (Moonshot) | API Key | Beta |
| Minimax | API Key | Beta |
| OpenAI Codex | OAuth | Beta |
| Gemini | — | Planned |
| OpenRouter | — | Planned |
Pace projection explained
The pace feature is the most useful part. It doesn't just show current usage — it projects whether you'll hit your limit:
- Headroom (green) — projected usage under 70% of quota
- On pace (green) — projected 70–100%, you're fine
- Over pace (red) — projected to exceed the limit before reset
This works for both 5-hour burst windows and 7-day rolling windows.
Installation
Download from GitHub Releases, run the installer, and it auto-detects your API keys. Configuration is stored locally in auth.json.
FAQ
How do I track my AI API spending?
AI Usage Tracker is a free open-source dashboard that monitors usage, costs, and quotas across multiple AI providers from a single compact window. It auto-detects API keys from environment variables and Claude Code credentials.
What AI providers can I monitor?
GitHub Copilot, Anthropic Claude Code, Mistral, DeepSeek, Kimi (Moonshot), Minimax, OpenCode Zen, Synthetic, Z.AI, and OpenAI Codex. Gemini and OpenRouter support is planned.