AgentTune
A neutral public price catalog for 50 AI agent platforms, with cost comparisons for real workloads.
Problem
The price of running an agent is split across vendor pages. Inference APIs, GPU clouds, serving engines, gateways, observability tools and agent frameworks each publish their own rates, in their own units. A cell filled from memory looks complete and is not a quote.
Solution
AgentTune is a public price catalog for 50 of those platforms. Every price is quoted from a public source page. When the page does not state a price, the field is left empty.
A compare view estimates cost per 1,000 runs for preset workloads, such as a coding agent, and shows where a dedicated GPU beats an API. Platform pages accept corrections through a moderated queue, and accepted changes land in a public changelog. A weekly refresh checks the source pages. It pairs with DevRadar.
Analysis
There are no sponsored rankings and no scores. A score would turn a quoted rate into a verdict about which platform is better, and a sponsored slot would sit in the same list as a number copied off a source page. The catalog does not name a winner. It shows the figure on the page, or that the page had none.