ExchangeRebel LLM Inference API

openapi: 3.2.0
info:
  title: ExchangeRebel — pay-per-call APIs for AI agents (x402 on…
  version: 1.0.0
  contact:
    name: 'ExchangeRebel (operator: Rebel Data)'
    url: https://exchangerebel.com
    email: sound.money.investing@gmail.com
  x-guidance: 'All paid operations use x402 on Solana: call without payment to get a 402 challenge (SOL recipient + USDC token account + fresh memo), pay with the memo attached via the Solana Memo program, then retry with header X-Payment-Tx: <signature> and ?toll_memo=<memo>. Challenges are dual-version: the PAYMENT-REQUIRED header carries the x402 v2 PaymentRequired object and the response body carries the same challenge in x402 v1 form for compatibility. Free operations (security: []) include each service''s /health catalog with full response shapes. Prices in x-payment-info are USD per call unless the summary says otherwise (RPC is per 1,000-request block; Telegram provisioning is per bot-month; transcription is per minute of audio).'
  description: '67 data and utility APIs behind x402 tolls, settled in SOL or USDC on Solana. Call a paid endpoint without payment to receive a 402 x402 challenge (recipient + fresh memo), pay with the memo attached via the Solana Memo program, then retry with the X-Payment-Tx header and toll_memo query parameter. Free catalog/health endpoints are marked without x-payment-info. Full catalogs: GET /toll/discovery and each service''s /health.'
servers:
- url: https://exchangerebel.com
tags:
- name: LLM Inference
paths:
  /toll/inference/chat:
    post:
      tags:
      - LLM Inference
      summary: 'Chat completion (GPT-OSS 120B): $0.01 flat per call'
      responses:
        '200':
          description: Success. Response shape is documented in the service catalog (GET /health on the service).
          content:
            application/json:
              example:
                choices:
                - finish_reason: stop
                  index: 0
                  message:
                    content: The capital of France is Paris.
                    role: assistant
                created: 1760880000
                id: chatcmpl-sample0001
                model: openai/gpt-oss-120b
                object: chat.completion
                usage:
                  completion_tokens: 8
                  prompt_tokens: 14
                  total_tokens: 22
        '402':
          description: 'Payment Required (x402). The PAYMENT-REQUIRED response header carries the x402 v2 PaymentRequired object (base64 JSON: scheme, CAIP-2 network, amount, asset, payTo, plus the bazaar discovery extension); the response body carries the same challenge in x402 v1 form for compatibility, naming the SOL recipient and USDC token account on Solana plus a fresh memo. To pay on this service''s rail: pay with the memo attached (Solana Memo program), then retry with header ''X-Payment-Tx: <transaction signature>'' and query parameter ''toll_memo=<memo>''.'
      requestBody:
        content:
          application/json:
            schema:
              type: object
              properties:
                body:
                  type: string
                  description: 'application/json (required) - model: string (required) - openai/gpt-oss-120b only; messages: array (required) - OpenAI chat messages; max_tokens: integer 1-500; temperature: number 0-2; stream must be absent/false'
              required:
              - body
      x-payment-info:
        protocols:
        - x402
        price:
          mode: fixed
          currency: USD
          amount: '0.01'
      security:
      - x402Payment: []
      operationId: postTollInferenceChat
      x-operation-id-source: derived
  /toll/inference/chat/preflight:
    post:
      tags:
      - LLM Inference
      summary: 'Free pre-check twin: exact $0.01 price before payment'
      responses:
        '200':
          description: Success. Response shape is documented in the service catalog (GET /health on the service).
      requestBody:
        content:
          application/json:
            schema:
              type: object
              properties:
                body:
                  type: string
                  description: application/json (required) - the same JSON body as POST /toll/inference/chat
              required:
              - body
      security: []
      operationId: postTollInferenceChatPreflight
      x-operation-id-source: derived
  /toll/inference/health:
    get:
      tags:
      - LLM Inference
      summary: Inference health (free, counts-only)
      responses:
        '200':
          description: Success. Response shape is documented in the service catalog (GET /health on the service).
      security: []
      operationId: getTollInferenceHealth
      x-operation-id-source: derived
  /toll/inference:
    get:
      tags:
      - LLM Inference
      summary: Inference catalog at the mount root (free)
      responses:
        '200':
          description: Success. Response shape is documented in the service catalog (GET /health on the service).
      security: []
      operationId: getTollInference
      x-operation-id-source: derived
components:
  securitySchemes:
    x402Payment:
      type: apiKey
      in: header
      name: X-Payment-Tx
      description: 'x402 toll payment proof: the Solana transaction signature of a payment carrying the memo from the 402 challenge; send together with ?toll_memo=<memo>. Payment itself is made on-chain in SOL or USDC to the recipient named in the challenge.'