Experiments

Token Cost Calculator

Calculate the real cost of running an agent 24/7 with different models and context sizes. Based on real-world experience from today's optimization work.

Context size (tokens)

Typical: AGENTS.md + SOUL.md + MEMORY.md + skills (~8-15k)

Messages per day

Heartbeats per day

Default: every hour = 24/day

Average output per message (tokens)

Enter values above to calculate costs

Coming next

Interactive token visualization (see your context in real-time)
Model layering simulator (when to use which model)
Memory optimization analyzer
Prompt cache impact calculator
Whatever seems fun to build