Qwen3.8 Max pricing on Vana API: $0.6944 per million

§ What Vana API is, in plain terms

Vana API is one address that answers API requests. That address is https://api.vanaapi.com. Everything you send goes there. The dashboard, where keys, balance and the usage log live, is on the main site, https://vanaapi.com. Those two places have different jobs, and knowing which is which is the only real setup step.

The selling point is simple: one address, three protocols, minimal setup. You do not juggle endpoints or study a long integration guide. You pick the model family you want, use the client that matches it, and point that client at the one address above. That is the whole shape of it.

This article is a plain introduction for newcomers, and it follows one concrete thread: reading the usage log in the Vana API dashboard after a session with Qwen3.8 Max. Along the way it will show what the log records, what the numbers mean, and how the prepaid balance works.

§ The three protocols on one address

Vana API speaks three protocols on the same address. Which one a call uses depends on the model family, and each family has its own client rules.

• Claude models: only from official Claude Code or Claude Desktop, version 2.1.285 or later, which speak Anthropic Messages (/v1/messages) themselves. They do not work from any other client, any SDK, cURL or a script.
• Gemini models: only from Gemini CLI, over the Gemini native protocol (/v1beta/models/...). They are not OpenAI-compatible.
• GPT, Grok, DeepSeek, GLM, Kimi, Qwen, MiniMax, MiMo: OpenAI format (/v1/chat/completions), from Cursor, Cline, Roo Code, OpenCode, Cherry Studio, any OpenAI SDK or cURL. GPT models also answer on /v1/responses, which is what Codex uses.

Qwen3.8 Max sits in that third group. Its model id is qwen3.8-max, and it speaks OpenAI format at /v1/chat/completions. So a session with it runs from Cursor, Cline, Roo Code, OpenCode, Cherry Studio, an OpenAI SDK or cURL. Nothing else is needed.

§ Reading the usage log after a Qwen3.8 Max session

Keys are created in the dashboard at https://vanaapi.com. Once a key exists and has balance, you run your session, and then you open the usage log. Every call is listed there. That is the part newcomers tend to appreciate: no estimating, no guessing at the end of the month. The log shows each call as it was charged.

Qwen3.8 Max is priced at $0.6944 / $2.08 per million tokens (input / output). Those two numbers are what the log is doing arithmetic with. Say, as an example, you run a session that reads 100,000 input tokens and writes 20,000 output tokens. Starting from the numbers above, 100,000 input tokens is a tenth of a million, so $0.6944 divided by 10 gives $0.06944. 20,000 output tokens is a fiftieth of a million, so $2.08 divided by 50 gives $0.0416. That example session comes to about $0.111. The usage log in the Vana API dashboard would show that call and that charge.

This is the everyday value of a per-token price you can read. Because billing is prepaid and charged per real token, with no monthly fee, the log is a complete record of what you spent and nothing more. Say you run this every day for a month: the log keeps a line per call, and the balance simply goes down as it goes.

§ What the balance looks like around it

Billing on Vana API is prepaid balance, charged per real token, with no monthly fee. You top up, you spend, and the usage log tells you exactly where the balance went. On sign-up there is a gift of $0.10 of credit, which is enough for a first small session with Qwen3.8 Max so you can see the log working before you commit anything.

There are also monthly plans, described here only as they are: Claude $1,000 / 30 days at $64.72 for 30 days, at most $320 a week, Claude models only; Claude $2,400 / 30 days at $153.61 for 30 days, at most $750 a week, Claude models only; and GPT $4,000 / 30 days at $103.61 for 30 days, at most $1,200 a week, GPT models only. Qwen3.8 Max is not in any of those plans, so a Qwen session runs on prepaid balance and appears in the usage log as described above.

Refunds follow one rule: unused balance can be refunded within 3 days of the top-up, minus what was used and the payment processor's fee. The referral programme pays 14% of what people you invite pay, and 5% one level further; earnings can be withdrawn from $13.89.

§ How the neighbours compare in the log

The usage log makes it easy to compare models on the same session shape, because every line is per real token. In the Qwen family, Qwen3.7 Plus (qwen3.7-plus) sits at $0.1278 / $0.4167 per million, while Qwen3.8 Max sits at $0.6944 / $2.08. Both speak OpenAI format from the same clients, so switching between them is a model id change and nothing more.

Other OpenAI-format models in the same group are MiniMax M3 (MiniMax-M3) at $0.1389 / $0.5556, MiMo V2.6 Pro (MiMo-V2.6-Pro) at $0.2083 / $0.3472, DeepSeek V4 Flash (deepseek-v4-flash) at $0.2778 / $1.04, DeepSeek V4 Pro (deepseek-v4-pro) at $1.11 / $3.33, GLM-5.3 (glm-5.3) at $0.5556 / $1.67, Kimi K3 (kimi-k3) at $1.04 / $5.21, and Grok 4.7 (grok-4.7) at $0.4861 / $1.53. Each of those is charged at its own per-token rate and shown as its own line in the usage log.

§ Setting it up once

Setup on Vana API is short. First, make a key in the dashboard at https://vanaapi.com. Second, give that key to the client that matches the model family you want, pointed at https://api.vanaapi.com. Third, run a session and read the usage log. For Qwen3.8 Max that means an OpenAI-format client at /v1/chat/completions, and the log will show the call priced at $0.6944 / $2.08 per million tokens.

That is the whole loop. One address, three protocols, and a log that accounts for every call. Whether you spend the $0.10 sign-up credit on a first Qwen3.8 Max session or move straight to prepaid balance, the picture in the dashboard stays the same: real tokens in, real tokens out, one line per call.