OmniProvider

OmniProvider: pay-per-call LLM API for AI agents

Chat completions and cited briefs, paid per request in USDC. No account, no API key, no subscription.

Your agent sends a request, receives a price (HTTP 402), pays on-chain, and gets the result. You are charged only after the result is produced.

OpenAPI description · llms.txt · Quick start · Contact

How it works

  1. Call a paid route with a JSON body.
  2. Receive a 402 challenge that states the exact price and where to pay.
  3. Pay with a standard agent payment client (x402 on Base, or MPP on Tempo).
  4. Get the result. Payment settles after the response is produced (settle-after-serve).

Two payment rails, one service

RailNetwork / assetRoutes
x402 v2 (exact)Base (eip155:8453), USDChttps://pay.omnidev.tech/x402/...
MPP (Tempo charge)Tempo mainnet (chain 4217), USDC.ehttps://pay.omnidev.tech/v1/...

One rail per URL. Using the other rail on a URL returns an error and you are not charged.

Services and prices (live today)

Live services and prices
ServiceRoutePrice per callLimits
Chat - nanochat/nano (MPP) / chat-nano (x402)$0.001up to 1,000 input / 256 output tokens
Chat - smallchat/small (MPP) / chat-small (x402)$0.002up to 8,000 input / 1,024 output tokens
Cited briefbrief$0.002a brief that cites only the quotes you supply

Prices are per call. The authoritative, machine-readable prices are in openapi.json; each 402 challenge also states the exact amount.

Built for agents

Quick start

N=$(openssl rand -base64 24 | tr '+/' '-_' | tr -d '=')
npx agentcash fetch https://pay.omnidev.tech/x402/chat-nano \
  --method POST --header 'content-type: application/json' \
  --header "X-Omni-Client-Nonce: $N" \
  --body '{"prompt":"hello"}' --payment-protocol x402

More examples, including MPP, are in the quick start.

Status and trust

Contact

Use the contact form. Please do not send secrets.