Frequently asked questions
Quick answers, dig into linked guides when you need code.
What request format does RodiumAi use?
OpenAI-compatible chat uses the JSON fields in POST /v1/chat/completions (base URL https://api.rodiumai.io/v1). Claude also works via Anthropic Messages at https://api.rodiumai.io. Start from the Quickstart for ready-to-copy examples.
Can I use the Anthropic SDK?
Yes. Point the official Anthropic client at https://api.rodiumai.io (no /v1, the SDK adds /v1/messages). See the Anthropic Messages guide and the Quickstart Anthropic section. Claude-first today; same RODI billing as OpenAI mode.
How do API keys look?
Keys created in the dashboard start with rd_sk_prod_ (or rd_sk_dev_ in development). Send Authorization: Bearer rd_sk_… on every route; x-api-key is accepted only on /v1/messages and /v1/audio/* (Anthropic SDK). Rotate keys from Dashboard → API keys when staff changes.
What is a RODI?
RODI is the internal prepaid unit for inference. Charges accrue based on tokens and published per-model rates in Pricing and debited from your wallet when each request settles. API responses report token counts in usage, never RODI amounts: see your spend in Dashboard → Usage, or compare GET /v1/wallet before and after a call.
Can I pay with Mobile Money?
Yes, top-ups are designed around the methods common across our launch markets. Purchases reconcile to your RODI balance for API usage.
How are rate limits applied?
Per API key and per model: requests per minute, tokens per minute and, for some models, requests per day, in fixed UTC windows. Each model's limits are in GET /v1/models (rodiumai_rate_limits); without one, 60 requests and 60,000 tokens per minute apply. Over the limit you get 429 with a Retry-After header in seconds. → Rate limits