Guides

RODI credits

Predictable metering for multi-model fleets — reconcile what you prepaid with what inference actually consumes.

Rodium abstracts cloud vendor invoices into prepaid RODI. Each model declares a transparent per-token (per 1K tokens) price — numbers live on Models marketing page and programmatically via GET /v1/models.

Formula

cost_rodi = (input_tokens + output_tokens) / 1000 × model_rate_per_1k

Worked example — assume Rodium Pro at 0.010 RODI / 1K tokens:

(200 prompt + 400 completion) ÷ 1000 × 0.010 = 0.006 RODI

Running balance after calls may appear alongside usage payloads—see responses in Chat completions reference.