$0/mo
Analyze 20 batch rows per month
- Unlimited single token calculations
- Unlimited API cost calculations
- 3 prompt optimizations per month
- 1 projects
Start free, then upgrade when AI cost management becomes a team workflow.
Single calculations stay free. Paid plans add batch rows, prompt optimization, projects, model price monitoring, and team cost reports. Paid API-powered prompt optimization is coming for Pro and Team plans.
$5/mo
Analyze 100 batch rows per month
- Unlimited single token calculations
- Unlimited API cost calculations
- 10 prompt optimizations per month
- 3 projects
$19/mo
Analyze 2,000 batch rows per month
- Unlimited single token calculations
- Unlimited API cost calculations
- 100 prompt optimizations per month
- 20 projects
- Model price alerts
$29/mo
Analyze 10,000 batch rows per month
- Unlimited single token calculations
- Unlimited API cost calculations
- 500 prompt optimizations per month
- 100 projects
- Model price alerts
- Team reports and 10 seats
Paid API-powered prompt optimization is coming for Pro and Team plans.
The free Prompt Optimizer will stay useful with local rules. Paid plans are being prepared for more precise official API-powered prompt rewrites, member quotas, usage history, and team reporting.
How to choose the right plan
Use Free when you only need quick token and API cost estimates. Move to Starter when you are analyzing real prompt files. Choose Pro when model price monitoring and multiple projects become important. Use Team when engineering, product, and finance need shared reports and exports.
Why usage limits are based on workflow value
Single calculators stay free because they help builders make better decisions before spending money. Paid plans focus on team workflows: bulk analysis, saved projects, prompt optimization, price alerts, and reporting.
Before you launch an AI product
Estimate best-case, normal-case, and heavy-use budgets before setting your own pricing. AI products often fail on margins not because the model is too expensive once, but because token usage compounds across users, retries, long context, and generated output.