# x402 Mesh Inference > Pay-per-prompt OpenAI-compatible chat completions routed across a mesh of independent GPU nodes, settled in USDC on Algorand via x402. ## Endpoint POST https://x402-mesh-gateway.fly.dev/v1/chat/completions Request and response bodies are OpenAI-compatible. Set `"stream": true` for a `text/event-stream` response terminated by the literal frame `data: [DONE]`. ## Paying Call /v1/chat/completions without a payment header to receive `402 Payment Required`, whose `accepts` array carries the exact requirements. Pay with an x402 client configured for the Algorand `exact` scheme; the facilitator sponsors the ALGO fee, so you sign only the USDC transfer. Your account must be opted in to the USDC ASA. ## Example ```json { "model": "llama3.1:8b", "messages": [ { "role": "system", "content": "You are a concise assistant." }, { "role": "user", "content": "Summarise the x402 payment protocol in one sentence." } ], "max_tokens": 128, "temperature": 0.2, "stream": false } ``` ## Live configuration These values are read from the running gateway at request time and override anything above if they differ. - endpoint: https://x402-mesh-gateway.fly.dev/v1/chat/completions - price: $0.0020 USDC per request (2000 atomic units, 6 decimals) - pricing: per-model; the price above applies to any model not listed below - llama-3.3-70b-versatile: $0.0060 USDC - network: algorand:SGO1GKSzyE7IEPItTxCByw9x8FmnrCDe (testnet) - asset: USDC ASA 10458941 - payTo: R5UN7S5G3AMKBDFEB42E7LSUT7JJGGZ6HL2S74ULBBZ36GWZVHEHJ53AOY - scheme: exact - facilitator: https://facilitator.goplausible.xyz - maxTimeoutSeconds: 180 - tags: x402-global-challenge, ai-inference, llm, openai-compatible, algorand - discovery: https://x402-mesh-gateway.fly.dev/.well-known/x402 - pricing: https://x402-mesh-gateway.fly.dev/v1/pricing - quickstart: https://x402-mesh-gateway.fly.dev/quickstart (runnable client, no wallet needed to read) - nodes: https://x402-mesh-gateway.fly.dev/v1/nodes - settlements: https://x402-mesh-gateway.fly.dev/v1/settlements