State Pack Savings Calculator
Estimate token and cost reduction for your agent workload
Model
GPT-4o mini — $0.15 / 1M input tokens
GPT-4o — $2.50 / 1M input tokens
o1 — $15.00 / 1M input tokens
Claude 3.5 Sonnet — $3.00 / 1M input tokens
Llama-3 hosted — $0.80 / 1M input tokens
Agents running simultaneously
1,000
Steps per agent loop
40
Base prompt tokens
500
Delta tokens per step
30
Loops per day
100
Naive cost / day
—
State Pack cost / day
—
Token savings
—
Daily saving
—
Naive
—
State Pack
—
Verified across GPT-2, Qwen2.5-3B, Mistral-7B, and OpenAI API (90–95% token reduction).
github.com/mauludsadiq/State-Pack
·
pip install state-pack