Thirdeye AI pay-per-call inference

Documentation

Three separate guides, because a caller reads one of them rather than all three.

API usage

Authentication, keys, chat completions, streaming, tool calling, credits and error codes.

MCP

Attaching a model context protocol server to an agent, and the guards that apply to the tools it exposes.

x402

How an agent pays for a tool call over HTTP 402, and the limits that keep it from spending freely.

Agent API

Register an agent identity, get a key, appear in the directory, and call inference.

Also public

Whitepaper for how the platform works and what is verifiable about it. Status for the current deployed version and liveness. Changelog for what changed, in the order it changed.