Docs | RDGlobal

Quick start

Create an account and save your recovery code.
Review service availability and add service credit when funding is activated.
Create a named API key with a monthly spend limit.
Use an ID returned by GET /v1/models.

Platform preview · Accounts and project requests are open. AI inference and card funding are awaiting activation.

Authentication

Send the key in the HTTP Authorization header using the Bearer scheme. A full key is shown only when it is created. Revoke keys from the console when they are no longer needed.

Authorization: Bearer $RDGLOBAL_API_KEY

Do not ship server keys in a mobile app, browser bundle or public repository. Browser-console sessions are separate from API-key authentication.

Available models

GET https://rdglobalrecy.com/v1/models

The response lists the models currently activated for the service, with pricing in USD per million input and output tokens. An empty list means no inference model is activated.

Create a chat completion

curl https://rdglobalrecy.com/v1/chat/completions \
  -H "Authorization: Bearer $RDGLOBAL_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "MODEL_ID",
    "messages": [
      {"role": "user", "content": "Explain this API."}
    ],
    "max_tokens": 256,
    "stream": false
  }'

model must be an enabled model ID. messages accepts 1–50 text messages with system, user or assistant roles. max_tokens must be within the published model maximum. This release supports non-streaming text responses; send stream: false.

Response fields

id is the RDGlobal request reference. choices contains generated output. usage includes prompt, completion and total tokens. billing.service_credit_charged gives the charged USD amount.

Duplicate protection

Include a unique Idempotency-Key header for a request. A repeated key for the same account is rejected with HTTP 409. Check usage before retrying a request with a new key if the original network response was interrupted.

Usage and errors

HTTP	Meaning
400	Unsupported or invalid request parameters
401	Missing, incorrect or revoked API key
402	Insufficient service credit or key spend limit
409	Previously submitted idempotency key
429	Request-rate limit exceeded
502	Upstream request could not complete; reservation released
503	Model or service not activated

Usage records contain model, token counts, charge, status, response time and timestamp. Prompts and generated answers are not retained in the RDGlobal usage ledger.

Managed compute

Use the compute section of the console to submit a workload specification. Review and accept the quote, then track the actual service activation. Compute provisioning is a managed process; the API does not automatically purchase virtual machines.

Your first integration.