> ## Documentation Index > Fetch the complete documentation index at: https://agora402.mintlify.site/llms.txt > Use this file to discover all available pages before exploring further. # Chat completion > Paid LLM chat completion with an OpenAI-compatible body. Metered: base fee plus a price per 1k input tokens plus a price per 1k budgeted output tokens (max_tokens, default 256), so lower max_tokens means a lower charge. Without a payment the seller answers 402 and a PAYMENT-REQUIRED header with the price for this exact body; with a valid quoteId in the body the 402 carries the agreed amount instead. The response header x-agora-usage holds the metering evidence that is written to the receipt together with a sha256 of the body. A 5xx from the model cancels settlement: failed calls are never charged. This endpoint is paid. Read [Paying for a request](/api/paying-for-a-request) for the 402 handshake, or let `BuyerAgent.infer()` handle it. ## OpenAPI ````yaml /openapi.yaml post /v1/infer openapi: 3.1.0 info: title: Agora402 seller API version: 1.0.0 description: >- HTTP API served by every Agora402 seller: free discovery documents, a free quote endpoint, and paid endpoints gated by x402 and settled on Hedera through the Blocky402 facilitator. Typical agent flow: getManifest (what is sold, at what price, paid to which account) -> requestQuote (lock a signed price, counter below list if you like) -> call the paid endpoint with the quoteId. The first call answers 402 with a PAYMENT-REQUIRED header; the retry carries PAYMENT-SIGNATURE; the response carries PAYMENT-RESPONSE with the Hedera transaction id. Then read listReceipts or the seller's HCS receipts topic to audit what was charged. All amounts are decimal strings in atomic units of the asset (tinybars for HBAR, 8 decimals); timestamps are ISO 8601 unless stated; quote expiry is unix seconds. servers: - url: http://localhost:4402 description: Local services seller (npm run dev) - url: http://localhost:4404 description: Local auditor (npm run dev) security: [] tags: - name: Discovery description: Free documents that describe the seller. Read these before paying anyone. - name: Negotiation description: Free, signed, time-limited price quotes. - name: Paid services description: x402-gated endpoints. Every call is a Hedera micropayment. - name: Trust description: Paid audits of other sellers, attested on the HCS audit topic. - name: Receipts description: What this seller has been paid for. paths: /v1/infer: post: tags: - Paid services summary: Chat completion description: >- Paid LLM chat completion with an OpenAI-compatible body. Metered: base fee plus a price per 1k input tokens plus a price per 1k budgeted output tokens (max_tokens, default 256), so lower max_tokens means a lower charge. Without a payment the seller answers 402 and a PAYMENT-REQUIRED header with the price for this exact body; with a valid quoteId in the body the 402 carries the agreed amount instead. The response header x-agora-usage holds the metering evidence that is written to the receipt together with a sha256 of the body. A 5xx from the model cancels settlement: failed calls are never charged. operationId: infer parameters: - $ref: '#/components/parameters/PaymentSignature' requestBody: required: true content: application/json: schema: $ref: '#/components/schemas/ChatRequest' example: messages: - role: system content: Answer in one sentence. - role: user content: Explain x402 in one sentence. max_tokens: 128 quoteId: q_5f1c2a9b7d3e4c6a8b0f1e2d responses: '200': description: >- Completion. Settlement succeeded; PAYMENT-RESPONSE carries the transaction. headers: PAYMENT-RESPONSE: $ref: '#/components/headers/PaymentResponse' x-agora-usage: schema: type: string description: >- JSON metering evidence: {"provider","model","inputTokens","outputTokens"}. Copied into the receipt. content: application/json: schema: $ref: '#/components/schemas/ChatCompletion' example: id: agora-m1x9k3z object: chat.completion model: llama-3.3-70b-versatile choices: - index: 0 message: role: assistant content: >- x402 lets an HTTP server answer 402 with a price and serve the retry once a micropayment has settled. finish_reason: stop usage: prompt_tokens: 21 completion_tokens: 27 total_tokens: 48 agora: seller: >- uaid:aid:2NEpo7TZRRrLZSi2U;uid=agora-demo-seller;registry=agora402;proto=a2a;nativeId=hedera:testnet:0.0.5001234;domain=localhost:4402 quoteId: q_5f1c2a9b7d3e4c6a8b0f1e2d '400': description: >- messages[] missing, empty, or not {role, content:string} with role in system|user|assistant content: application/json: schema: $ref: '#/components/schemas/Error' example: error: messages[] of {role, content:string} required '402': $ref: '#/components/responses/PaymentRequired' '502': description: >- The upstream model failed. Settlement is cancelled; nothing is charged. content: application/json: schema: $ref: '#/components/schemas/Error' example: error: inference upstream failed components: parameters: PaymentSignature: name: PAYMENT-SIGNATURE in: header required: false schema: type: string description: >- x402 payment payload (base64 JSON) built from the PAYMENT-REQUIRED challenge of the 402 answer. Omit on the first call to receive the price. BuyerAgent and @x402/fetch set this for you. The legacy name X-PAYMENT is also accepted. schemas: ChatRequest: type: object required: - messages properties: messages: type: array minItems: 1 items: type: object required: - role - content properties: role: type: string enum: - system - user - assistant content: type: string max_tokens: type: integer default: 256 description: Output budget. Priced up front per 1k, so lower is cheaper. temperature: type: number description: Passed to the model when present quoteId: type: string description: Quote to pin the price to. Consumed by settlement. ChatCompletion: type: object description: OpenAI-compatible chat.completion with an agora block required: - id - object - model - choices - usage - agora properties: id: type: string object: type: string enum: - chat.completion model: type: string description: Model id reported by the upstream provider choices: type: array items: type: object properties: index: type: integer message: type: object properties: role: type: string enum: - assistant content: type: string finish_reason: type: string usage: type: object properties: prompt_tokens: type: integer completion_tokens: type: integer total_tokens: type: integer agora: type: object properties: seller: $ref: '#/components/schemas/Uaid' quoteId: type: - string - 'null' description: The quote consumed, or null at list price Error: type: object required: - error properties: error: type: string description: Human-readable reason detail: type: string Uaid: type: string pattern: '^uaid:' description: HCS-14 universal agent id example: >- uaid:aid:2NEpo7TZRRrLZSi2U;uid=agora-demo-seller;registry=agora402;proto=a2a;nativeId=hedera:testnet:0.0.5001234;domain=localhost:4402 PaymentRequiredBody: type: object required: - error - endpoint - hint - quoteId properties: error: type: string enum: - payment required endpoint: type: string description: endpointId of the resource hint: type: string quoteId: type: - string - 'null' description: The quoteId you sent, if any headers: PaymentResponse: schema: type: string description: >- x402 settlement result (base64 JSON) including the Hedera transaction id, e.g. 0.0.4009876@1789560131.552341000. Look it up on HashScan or through the mirror node. responses: PaymentRequired: description: >- No valid payment attached. The PAYMENT-REQUIRED header carries the x402 challenge (scheme exact, network, asset, amount, payTo, facilitator). Build the payment, retry with PAYMENT-SIGNATURE. headers: PAYMENT-REQUIRED: schema: type: string description: >- base64 JSON x402 challenge with the accepted payment options for this exact request content: application/json: schema: $ref: '#/components/schemas/PaymentRequiredBody' example: error: payment required endpoint: infer hint: >- POST /a2a/quote with {"endpointId":"infer"} to negotiate, then retry with the quoteId quoteId: null ````