AI Cost Lab
Monthly AI Cost Planner

Turn product usage into an AI API budget.

Estimate request volume, retries, cached input, per-user cost and annual spend with explicit low, expected and high assumptions.

Pricing verified See formulas and assumptions

Product usage

Start with the inputs your product team can reasonably estimate.

USD official list prices
No exchange-rate assumptions are introduced.
Advanced usage and pricing assumptions
Transparent scenarios

Low, expected and high usage

Only request volume changes. Token mix, retry rate, caching and pricing stay the same.

Scenario Visible assumption Monthly requests Monthly cost Annual cost
Low 70% of expected volume 0 $0.00 $0.00
Expected 100% of entered volume 0 $0.00 $0.00
High 150% of expected volume 0 $0.00 $0.00

Formula

successful requests = users × operations/day × active days

retry requests = successful requests × retry rate

monthly cost = cost/request × (successful + retry requests)

Cached input is separated from uncached input using the official cached-input rate where one is published. Missing price types are not silently treated as zero.

What to validate before budgeting

  • Use provider-reported token counts when available.
  • Measure actual retry and failure rates after launch.
  • Check whether batch latency fits the workload.
  • Validate model quality, context and tool support before switching to a cheaper alternative.

Last pricing update: .

Per-request calculator

Model one representative API call with cache, tools and long-context pricing.

Open →

Compare all active models

Review normalized request and monthly costs across the registry.

Compare →