Sticky utility · Founder tool

AI Token Burn & Runway Calculator

How many months before the money runs out, once every prompt has a price? Payroll, GPUs, subscriptions and tokens in one model, in pounds or dollars.

Company

Cash in bank (£)

Fixed monthly burn (£)Payroll, rent, legal, everything except AI costs.

GPU and hosting per month (£)

AI tooling subscriptions (£)Copilots, eval platforms, observability, vector database plan.

Customers

Monthly active users today

User growth per month (%)

Revenue per user per month (£)

Usage

Requests per user per day

Input tokens per requestSystem prompt, retrieved context and conversation history.

Output tokens per requestInclude reasoning tokens: they are billed as output.

Agent loop multiplierModel calls per user action: 1 for chat, 3 to 10 for agents with tool use.

Model pricing (USD per million tokens)

Main model, input

Main model, output

Small model, input

Small model, output

GBP/USD exchange rateAPI bills are usually in dollars. Used to convert when you work in pounds.

Levers

Traffic routed to small model (%)

Input served from prompt cache (%)

Cache discount on input (%)

Token bill, month 1…

Inference per user…

AI gross margin…revenue less tokens and GPUs

Runway…

Month 1 net burn: (). Shareable link ·

Cash balance

Projected over 36 months. Hover for monthly values.

Monthly revenue against total cost

Revenue Total cost (incl. tokens)

Month by month

Month	Users	Revenue	Tokens	Total cost	Net	Cash

How the model works Cost per request = (uncached input × input price + cached input × input price × (1 − cache discount) + output × output price) ÷ 1,000,000, blended between the main and small model by your routing share, then multiplied by the agent loop factor. Users compound monthly at your growth rate; revenue scales with them; fixed costs stay flat. Default prices are illustrative, so check your provider’s current price list. Your inputs stay in this browser. For the reasoning behind the levers, read the AI margin teardown.