ttop for your AI costs
Real-time terminal monitoring of LLM token usage and spending across providers and coding agents.
brew install tokentopapp/tap/tokentop 
Blind spots in your AI spending
You're coding with AI agents all day. Claude Code, OpenCode, Cursor â they're burning through tokens and you have no idea how fast. You check the Anthropic dashboard. It's hours behind. You check OpenAI. Different dashboard, different login. Want to see it all in one place? In real time? In your terminal, where you already live?
Scattered dashboards
Different login for every provider. Anthropic, OpenAI, Google â none talk to each other.
Hours behind
Provider billing dashboards update slowly. By the time you see the damage, it's done.
No spending limits
Most providers don't let you set daily caps. One runaway agent can burn through hundreds.
No local visibility
Your terminal is where you work. Why should cost monitoring live in a browser tab?
Everything you need to track AI spending
Built for developers who want granular control and visibility over their AI infrastructure costs.
Real-time Dashboard
Live token counts, costs, burn rate, and activity sparklines updating every second.
Session Tracking
Every coding agent session logged with model, tokens, cost, and duration.
Budget Guardrails
Set daily, weekly, or monthly spending limits with visual warning indicators.
Smart Sidebar
Break down spending by model, project, or agent type instantly.
Efficiency Insights
Analyze cache leverage, output verbosity, and cost-per-request metrics.
Historical Trends
Visualize cost patterns over 7, 30, or 90 days with ASCII step charts.
Provider Limits
Monitor rate limits and quota usage with visual gauge indicators.
15 Built-in Themes
Tokyo Night, Dracula, Nord, Catppuccin, and more to match your terminal.
Plugin System
Extend with custom providers, agents, themes, and notification hooks.
Zero Config
Auto-discovers credentials from Claude Code, env vars, and CLI auth files.
Works with your stack
Supported Providers
Supported Agents & CLIs
Plugin Registry
33 built-in plugins ready to use.
Anthropic Claude subscription usage tracking via OAuth.
OpenAI API usage tracking via OAuth. Supports GPT-4.1, o3, and all current models.
Gemini CLI usage tracking via Google OAuth.
GitHub Copilot usage tracking including premium requests.
OpenAI Codex subscription usage tracking via OAuth.
Perplexity AI API usage tracking (credit-based).
Antigravity (Google Gemini Advanced) subscription usage tracking.
MiniMax usage and quota tracking.
OpenCode Zen â curated AI models for coding agents.
Z.ai coding plan usage and quota tracking.
OpenCode session parsing, credential discovery, and real-time cost tracking.
Claude Code session parsing, credential discovery, and real-time cost tracking.
Cursor session parsing, credential discovery, and real-time cost tracking.
Windsurf (Codeium) session parsing, credential discovery, and cost tracking.
Gemini CLI session parsing, credential discovery, and real-time cost tracking.
Antigravity session parsing, credential discovery, and real-time cost tracking.
Simple terminal bell (BEL character) for budget alerts.
Visual screen flash using ANSI escape sequences for alerts.
plugin-sdk
SDK for building tokentop plugins â type definitions, helpers, and test harness.
See it in action
Watch tokentop monitor AI sessions in real-time.
Try it yourself â run ttop demo for a live preview with sample data.
Built in the open
tokentop is 100% open source. Join us in building the best terminal dashboard for AI engineering.