Alibaba

Qwen4-Max

Alibaba's new flagship for coding and knowledge work, with room for long source material and adjustable reasoning.

Adjustable reasoningtext inputtext output

Replaces Qwen 3.8 Max: input $2.1 → $1.575 per 1M (-25%), output $6.3 → $4.725 per 1M (-25%), same context window, max output 128K → 384K.

Run this model

alibaba/qwen4-max

Use an endpoint supported by this model.

Get a key for this setup →
Thinking guide →
const response = await fetch('https://api.minirouter.sh/v1/chat/completions', {  method: 'POST',  headers: {    Authorization: `Bearer ${process.env.MINIROUTER_KEY}`,    'Content-Type': 'application/json',  },  body: JSON.stringify({    model: 'alibaba/qwen4-max',    messages: [{ role: 'user', content: 'Why is the sky blue?' }],    max_tokens: 1024,  }),}) const data = await response.json()console.log(data.choices[0].message.content)
Read docs →

MiniRouter supports text input. Start with a small text task; reasoning can reserve the model's full output allowance even when you request a short answer.

Model overview

Input / 1M
$1.575
All-in MiniRouter rate
Output / 1M
$4.725
All-in MiniRouter rate
Context
1M
1,000,000 tokens
Max output
384K
tokens
Reasoning
low · high · xhigh
Choose the effort for the task
Workflows
Code · documents · tools
Your application runs and checks tools
Announced
15 Nov 2026
publisher release

Compare text-model alternatives or estimate token costs for your workload.

Choose your first task

Bring the evidence, then ask for a decision.

A large context window is useful when the answer depends on several files or documents. Give the model a defined question and enough source material to justify its answer.
  1. [ 01 ]

    Review a change

    Supply the relevant code, requirements and test output. Ask for concrete failure cases and a proposed fix.

  2. [ 02 ]

    Reconcile documents

    Label each text source and ask for conflicts, supporting passages and unresolved questions.

  3. [ 03 ]

    Build a tool loop

    Provide narrowly scoped tools, return their results to the model, and check the result before allowing consequential actions.

Spend effort deliberately

Compare effort on your own acceptance cases.

The catalog offers a reasoning toggle, low/high/xhigh effort and a token budget. More thinking is not a substitute for a clear task or a test that can reject a wrong answer.
  1. [ 01 ]

    Start with a bounded task

    Try a single review or extraction before handing over a whole workflow. Record correctness, elapsed time and billed usage.

  2. [ 02 ]

    Change one control

    Compare effort settings using the same input and acceptance criteria. Avoid combining an effort setting with a separate thinking budget.

  3. [ 03 ]

    Keep limits explicit

    Context and maximum output are ceilings, not target request sizes. Reasoning and the visible answer both contribute to cost.

From Qwen3.8-Max

Retest the workflow, not just the model name.

Qwen4-Max is the next flagship in the Max family. Keep Qwen3.8-Max as a comparison on your existing tasks, especially where prompts depend on a particular answer style.
  1. [ 01 ]

    Use the exact model ID

    Select alibaba/qwen4-max. A saved alias or router may still select a different model.

  2. [ 02 ]

    Keep inputs textual

    Convert source documents to text before sending them. Image and video inputs are not available on this MiniRouter route.

  3. [ 03 ]

    Validate outputs in your application

    Do not rely on schema-enforced JSON for this route. Parse and validate any machine-readable answer before using it.

Route

One route, standard clients.

Authenticate with a MiniRouter bearer key. Caller-provided upstream credentials are not used on this route.
Base URLhttps://api.minirouter.sh/v1
Model IDalibaba/qwen4-max
Endpoints/v1/chat/completions/v1/messages/v1/responses
Parametersmax_tokens, temperature, stop, tools, tool_choice, reasoning, include_reasoning
Structured outputNot enforced; a requested format is dropped and the answer is plain text.
ModalitiesInput: text · Output: text

Configured providers. Priority follows the live catalog order.

ProviderPriorityInput / 1MOutput / 1MParametersAccess
vercel
1$1.575$4.725max_tokens, temperature, stop, tools, tool_choice, reasoning, include_reasoningMiniRouter balance

Every model, one key.