AI//COST
Research AI Tools
AI//COST is an independent directory that tracks source-verified API pricing, with calculators and side-by-side comparisons across hundreds of models.
What does AI//COST do?
AI//COST is a curated pricing desk for AI model users who need reliable, source-verified API costs. It maintains a live index of per-model pricing for 288 models across 28 providers, updated with verification dates so you can see what changed and when.
Use the Pricing Hub to explore per-model rates and historical price changes, or run the built-in Calculators to estimate LLM API cost and token usage, plus subscription break-even and TTS pricing on live rates. For decision-making, the Compare section provides source-verified head-to-head comparisons with benchmarks and use-case verdicts.
AI//COST also supports deeper evaluation with Capability Q&A pages that answer questions like “Can X do Y?” using verified test outputs and screenshots. The directory is designed to avoid sponsored influence—no paid placements and no sponsored rankings—so the cost and comparison workflow stays focused on evidence.
What does AI//COST track—model costs only, or more than pricing?
AI//COST focuses on source-verified API pricing, including per-model rates, historical price changes, and blended effective-rate assumptions. It also includes capability Q&A and comparisons with verified outputs and benchmarks.
How are the pricing numbers verified?
Pricing data is verified manually against each vendor’s canonical pricing page, with a verification date stamped next to the published numbers.
What calculators are available on AI//COST?
The live calculators cover LLM API cost and token counting, plus subscription break-even and TTS pricing—using the latest verified rates.
What does “blended rate” mean on the pricing pages?
Blended rate uses a fixed workload assumption (92% input / 8% output, with an 82% cached workload) to compute an effective figure. Raw input and output rates reflect vendor-published numbers, and no cached discount is applied when a vendor doesn’t provide a cached-input rate.
Does AI//COST run paid placements or sponsored rankings?
No—paid placements and sponsored rankings are not used. Sponsored slots, where present, are visually marked and disclosed.
Is there a way to compare two models for my use case?
Yes. Use the Comparisons section for source-verified live pricing and benchmarks, with a verdict by use case.