New · 2026-11-16

Qwen4 Max API

Build with Qwen’s latest flagship. One API key for your code, multimodal workflows and agents.

Pay as you go · One key, three API formats
Get $5 credit. Pay $3.Save $2.00 on your first top-up

Built for the whole task

1M context

Bring up to one million tokens into a single conversation.

Up to 262K output

Generate up to 262K tokens in thinking mode, or 131K without thinking.

Text, images + video

Send text and visual inputs, and get a text response.

Tools + JSON Schema

Use parallel function calls and structured JSON outputs.

Work out your API cost

Monthly cost calculator

Model
90%
of input served from cache

Estimated monthly costQwen4 Max
Official$11.18
BetterToken$8.05

You save $3.13 per month

Get API key

Prices per 1M tokens. Input includes cache hits. Sources: BetterToken and Alibaba Cloud Model Studio · checked 17 Nov 2026.

Top up once. Get more from every token.

Choose your balance credit. The discount comes off what you pay.

Effective price per 1M tokens after this offer
Balance creditYou payYou saveQwen4 Max input / output
$50You pay: $48 · You save: $2$48$2$1.0368 / $3.1104Top up
$100You pay: $95 · You save: $5$95$5$1.026 / $3.078Top up
$200You pay: $185 · You save: $15$185$15$0.999 / $2.997Top up
$500Lowest tier rateYou pay: $450 · You save: $50$450$50$0.972 / $2.916Top up

Qwen4 Max vs Qwen3.8 Max. One key.

The same balance also works with Claude, GPT, Gemini and more.

per million tokens

Qwen4 Max

Input
$1.08−28%vs official$1.50
Output
$3.24−28%vs official$4.50
Cache read
$0.054−28%vs official$0.075
Context
1M
Max output
131K / 262K
non-thinking / thinking
Input
Text · images · video
API formats
Chat Completions
Anthropic Messages
Responses

Qwen3.8 Max

Input
$1.44−28%vs official$2.00
Output
$4.32−28%vs official$6.00
Cache read
$0.18−28%vs official$0.25
Context
1M
API formats
Chat Completions

Alibaba reports about 60% lower average cost per task than Qwen3.8 Max in its own tests.

Put Qwen4 Max to work

Copy the setup for your AI agent, or start with an SDK example.

1. OpenAI Chat Completions

curl https://bettertoken.ai/v1/chat/completions \
  -H "Authorization: Bearer YOUR_BETTERTOKEN_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "qwen4-max",
    "messages": [{"role": "user", "content": "Hello!"}]
  }'

2. Anthropic Messages

curl
curl https://bettertoken.ai/v1/messages \
  -H "x-api-key: YOUR_BETTERTOKEN_KEY" \
  -H "anthropic-version: 2023-06-01" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "qwen4-max",
    "max_tokens": 1024,
    "messages": [{"role": "user", "content": "Hello!"}]
  }'
3. OpenAI Responses
curl
curl https://bettertoken.ai/v1/responses \
  -H "Authorization: Bearer YOUR_BETTERTOKEN_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "qwen4-max",
    "input": "Hello!"
  }'

A few things you might want to know

Per million tokens, BetterToken charges $1.08 for input, $3.24 for output and $0.054 for cache reads — 28% below the official Qwen4 Max API price of $1.50, $4.50 and $0.075. Thinking tokens are billed as output. Your first top-up of $5 costs $3.

Create a BetterToken account or sign in, create an API key in Workspace, then top up your balance. Use qwen4-max as the model name in your first request.

No. Sign up with BetterToken, create an API key and call qwen4-max with it. You pay from your BetterToken balance — no Alibaba Cloud account or separate contract is needed.

Qwen4 Max has lower token prices: $1.08 / $3.24 for input / output, compared with $1.44 / $4.32 for Qwen3.8 Max. Alibaba reports about 60% lower average task cost in its own tests. Compare them on your workload. Qwen3.8 Max currently supports Chat Completions only.

Alibaba Cloud Model Studio's international list price for Qwen4 Max: $1.50 input, $4.50 output and $0.075 cached input per million tokens, checked on 17 Nov 2026. Our price is 28% lower on each of the three.

Qwen4 Max has a 1,000,000-token context window. Maximum output is 131,072 tokens without thinking, or 262,144 with thinking enabled.

Thinking is enabled by default and thinking tokens are billed as output. Set enable_thinking: false to turn it off. reasoning_effort accepts low, high or xhigh; the default is high. A lower effort setting does not guarantee lower cost.

Yes. We tested tool calls (function calling), JSON Schema structured output and streaming through Chat Completions on BetterToken. Alibaba also documents parallel function calls for Qwen4 Max.

Qwen4 Max supports OpenAI Chat Completions, OpenAI Responses and Anthropic Messages. Configure the OpenAI Python or Node SDK with https://bettertoken.ai/v1. For Anthropic, use https://bettertoken.ai and an x-api-key header.

Yes. Qwen4 Max accepts text, images and video and returns text. Send media as base64 or a public URL that can be downloaded directly.

Input tokens served from cache are billed at $0.054 per million instead of $1.08. The cache-read price applies only to tokens the response reports as cached, so repeated long prompts can cost less.

Yes. The same BetterToken balance also works with Qwen3.8 Max, Claude, GPT, Gemini and other models — set the model name in each request. See all prices on Models & pricing.

No subscription is required to use the Qwen4 Max API. Top up your balance and pay only for the tokens you use.

Top up in US dollars with Stripe or USDT on TRC-20. On your first top-up, get $5 credit for $3. See the table above for regular top-up offers.

YOUR NEXT PROJECT STARTS HERE

Build your next project with Qwen4 Max.

Your first $5 of API credit costs $3.