Inference.town turns model routing into a small-town utility

Field notes

The homepage lays out a developer workflow around an OpenAI-compatible /v1/chat/completions endpoint: choose a model, supply a bearer key, and send familiar chat JSON, with streaming shown in the example. Its model-and-pricing “market board” sits alongside usage, provider, funding, and API-key sections. The page states that inference can be routed across providers according to cost and that payment can happen per call through x402 or through a credit balance.