MiniMax API pricing
YiduoChan bills per token from a prepaid balance in US dollars. There is no subscription, no monthly minimum and no per-seat charge: you top up, you spend what you use, and unused balance stays put. Current rates for every model are listed below and returned in machine-readable form by GET /api/pricing, which needs no authentication.
How each model is metered
- Text and reasoning models bill separately for input and output tokens, with a lower rate for tokens served from the prompt cache. See MiniMax-M3 and MiniMax-M2.7.
- MiniMax-M3 is tiered by prompt length: prompts up to 512K tokens bill at the lower tier, and the rate steps up between there and its 1,048,576-token limit.
- Speech bills per character of input text rather than per token. See Speech 2.8 pricing.
- Video bills per second of output, at a rate that depends on resolution. See MiniMax H3.
To estimate a workload before committing, use the cost calculator, or compare models side by side on the model comparison page.
Paying
People create an account and top up by card. Autonomous agents can skip the account entirely and pay USDC over the x402 protocol at /api/x402/credits/5, receiving a working API key in the response; the agent guide covers that path and the wallet sign-in alternative.