Mira has launched Sterling, an AI inference service that remains in testing and lets developers send model requests through screened providers while paying according to use.
An API key lets developers connect their software to Sterling’s endpoint. Its current menu lists Claude, GPT and Gemini.
Before routing traffic, Sterling checks whether providers serve the specified model version and handle tool calls; requests go to providers that pass.
Launch prices are up to 50% below list prices. Payments can be made in USD Coin (USDC) through x402 or in MIRA.
Sterling brings AI inference into the range of services using x402, a protocol that lets a website or API request payment as part of a web transaction. EconoTimes’ earlier count of Base x402 activity recorded more than 138 million transaction events in the 12 months through July 1, 2026, across different services. The count did not identify customers or agents and gave no share for AI inference.
