Life of a request

Six beats from POST to receipt: auction, 402, payment, stream, meter, refund. Chain latency never touches your tokens.

Everything Omnious does happens inside one HTTP exchange. Here is the whole flow, then each beat in detail.

1 · POST

You send an OpenAI-compatible chat request to the router. No API key is required; the payment that follows is the authentication.

2 · The auction

The router scores every live quote for your model class in memory: price, measured latency, and quality, weighted by your priority. This takes under a millisecond and adds no network round trip. The winner and the cleared price are fixed here, before any money moves. Details in Second-score clearing and Scoring.

3 · 402 Payment Required

The router answers with the one HTTP status code the web never used: 402. The body quotes your worst-case cost at the winner's cleared prices (your token budget times the cleared rates). This is an upper bound, not the bill.

4 · Sign and retry

Your client signs a USDC authorization for that maximum and retries the request with the x402 v2 payment header. The router verifies the signature and balance in about 100 ms. Gas is sponsored; you only ever hold USDC.

5 · Stream, meter, settle concurrently

Inference starts immediately on verification. On-chain settlement runs beside the stream, never in front of it, so chain latency cannot touch your time-to-first-token. The reply streams through a metering proxy that counts tokens and measures the provider's real latency, which feeds future auction scores. Bad-debt exposure from this optimistic start is capped per wallet and router-wide.

6 · Receipt and refund

You are charged metered actuals, not the authorized maximum, and any overage refunds automatically. The signed receipt records token counts, cleared prices, and payment references, and each hour's records are hashed and anchored on-chain. The full economic breakdown (the reconstructed auction, every rival's counterfactual price) is one call away. See Receipts & analytics.