Components
Token Usage
Monitor token consumption, context window utilization, and session costs across models
Demo
Click the terminal below and press space to start simulating API requests.
Features
- Context window bar with color-coded breakdown of system, history, RAG, and response tokens
- Session budget tracker showing cost against a configurable limit
- Multi-model support for comparing token economics across Claude Sonnet, Haiku, GPT-4o, and GPT-4o mini
- Live request simulation with token counts, costs, and latency stats
Keybindings
spacestart/stop simulationleft/rightswitch modelrreset session