MiniMax M3 pay-as-you-go pricing on Vana API, $0.1389 per 1M in

When a billing question comes up, the fastest way to settle it is to run one small workflow from start to finish and watch the numbers appear where they belong. This article does exactly that. Everything below is an example, built only from figures that Vana API publishes, so you can follow the same shape with your own calls.

The idea underneath is simple: one address, three protocols, minimal setup. For MiniMax M3 that means a single endpoint, the OpenAI format, and nothing more to install than the client you already use. Vana API keeps the rest of the work in the dashboard, where keys, balance and the usage log live.

§ The one address and the protocol that fits MiniMax M3

Start with the endpoint. API requests go to https://api.vanaapi.com, and that address answers API requests only. The dashboard sits on the main site, https://vanaapi.com, and that is where you create keys, check your balance and read the usage log. Keeping the two apart is deliberate, and it is the reason setup stays short.

Three protocols are served from that one address. Claude models are spoken to only from official Claude Code or Claude Desktop, version 2.1.285 or later, which speak Anthropic Messages (/v1/messages) themselves. Gemini models are reached only from Gemini CLI, over the Gemini native protocol (/v1beta/models/...). GPT, Grok, DeepSeek, GLM, Kimi, Qwen, MiniMax and MiMo models use the OpenAI format (/v1/chat/completions), from Cursor, Cline, Roo Code, OpenCode, Cherry Studio, any OpenAI SDK or cURL.

MiniMax M3 sits in that third group. Its model id is MiniMax-M3, and it answers on /v1/chat/completions. That is the whole protocol decision for this example. No second client, no second address, no second set of credentials.

§ Setting up a key and a prepaid balance

Keys are created in the dashboard on https://vanaapi.com. The balance is prepaid, and calls are charged per real token, with no monthly fee. Every call is listed in the usage log in the dashboard, which is where the billing question in this example will be answered.

On sign-up there is a gift of $0.10 of credit. For this example, say you top up and then run a short batch of MiniMax M3 calls, then open the usage log to read what was spent. The log shows each call, so you can match the line items against the prices below rather than guess at them.

If you decide the balance is not needed, unused balance can be refunded within 3 days of the top-up, minus what was used and the payment processor's fee. That rule is worth knowing before a first top-up, because it sets the window for the question this article is walking through.

§ What MiniMax M3 costs on Vana API

Here are the published prices for this model. MiniMax M3 (model id MiniMax-M3) is $0.1389 / $0.5556 per million tokens on Vana API, input first and output second. Those two numbers are the ones to read off the usage log when you reconcile a bill.

For a worked example, say you run this every day for a month and your calls use 2,000,000 input tokens and 500,000 output tokens in total. Starting from the published figures, 2,000,000 input tokens is 2 x $0.1389, which is $0.2778, and 500,000 output tokens is 0.5 x $0.5556, which is $0.2778. Added together that is $0.5556 for the month. This is an example, not a quote, and your own totals will follow the same multiplication.

The same arithmetic works at smaller scale. Say you send 100,000 input tokens and 40,000 output tokens in one session. That is 0.1 x $0.1389, which is $0.01389, plus 0.04 x $0.5556, which is $0.022224. The session totals $0.036114. Again, this is an example built from the published prices.

§ Reading the usage log to answer the billing question

The usage log in the dashboard lists every call. For this example, open it after the session above and look for the lines that used MiniMax-M3. Each line carries the tokens that were charged, so you can add the input tokens and the output tokens yourself and compare the result with the figures you calculated from $0.1389 / $0.5556.

That is the whole reconciliation. Because billing is per real token and there is no monthly fee, there is no base charge to subtract and no allowance to track. The log is the record, and the balance moves by exactly what the calls cost.

Vana API keeps this consistent across the other OpenAI-format models as well, so a billing question about one model resolves the same way as a billing question about another. The difference is only the price line you read off the list.

§ Where monthly plans fit, and where they do not

Monthly plans exist for Claude models and for GPT models only. The Claude $1,000 / 30 days plan is $64.72 for 30 days, at most $320 a week, Claude models only. The Claude $2,400 / 30 days plan is $153.61 for 30 days, at most $750 a week, Claude models only. The GPT $4,000 / 30 days plan is $103.61 for 30 days, at most $1,200 a week, GPT models only.

MiniMax M3 is not covered by any of those plans, so for this model the prepaid balance is the path. That keeps the example simple: one address, one protocol, one price line, one log to read.

§ What the example leaves you with

From start to finish, the workflow was: create a key in the dashboard on https://vanaapi.com, top up a prepaid balance, send MiniMax M3 calls to https://api.vanaapi.com on /v1/chat/completions, then read the usage log to confirm the charge. Four steps, and the only figures you need are the published prices.

That is what one address, three protocols and minimal setup means in practice. Vana API handles the routing, the dashboard holds the keys, balance and usage log, and the billing question resolves to arithmetic you can check yourself.

There is also a referral arrangement alongside this: 14% of what people you invite pay, 5% one level further, and earnings can be withdrawn from $13.89. It is separate from the per-token billing, and it does not change how the usage log reads.