# Qwen4 Max Preview: Qoder and Token Plan Access

> Find where Qwen4 Max Preview is available, how Qoder and Token Plan access differ, and which preview limits matter before you start a coding task.

- Canonical HTML: https://omniakey.com/blog/qwen4-max-preview
- Markdown: https://omniakey.com/blog/qwen4-max-preview.md
- Published: 2026-11-11 (updated 2026-11-11)
- Author: Kiki, COO at OmniaKey
- Language: en
- Category: Guide
- Reading time: 5 min
- All posts: https://omniakey.com/blog/llms.txt

---

As of November 11, 2026, Qwen4-Max-Preview is available in Qoder's native selector and through Token Plan API access at Alibaba Cloud Model Studio in Beijing or QwenCloud in Singapore. The published Token Plan preview ends November 25 at 24:00 UTC+8; Qoder's native listing shows no expiry.

**Short answer**

- Qwen4-Max-Preview is Alibaba's preview multimodal model for coding, office productivity, frontend design, and complex agentic tasks, with the model ID `qwen4-max-preview` ([Model Studio](https://help.aliyun.com/en/model-studio/qwen4-max-preview), [QwenCloud](https://docs.qwencloud.com/developer-guides/getting-started/latest-model)).
- Qoder's native selector uses Qoder Credits, and availability comes from the account selector. The QwenCloud Token Plan connection requires an eligible Qoder tier.
- API access is Token Plan only. Model Studio and QwenCloud are separate plans; keep the account, key, and host together.
- Limits: a 1M-token context and a 393,216-token maximum output shared by thinking and the answer ([Model Studio](https://help.aliyun.com/en/model-studio/qwen4-max-preview)).

## Where can you use Qwen4-Max-Preview today?

Qwen4-Max-Preview has a native Qoder entry and published Token Plan access. The table keeps the access choices and bills separate.

|Access|Who can use it|Billing|
|-|-|-|
|Qoder native selector|Qoder account; selector decides availability|Qoder Credits at the displayed rate|
|Model Studio Token Plan (Beijing)|Model Studio Personal or Team plan|Model Studio Token Plan Credits|
|QwenCloud Token Plan (Singapore)|QwenCloud Individual or Team Edition Token Plan account|QwenCloud Token Plan Credits|
|Pay-as-you-go API|Not offered for `qwen4-max-preview`|—|

For the QwenCloud connection inside Qoder, the eligible tiers are Pro Trial, Pro, Pro+, or Ultra; Free and Teams are excluded ([QwenCloud Qoder guide](https://docs.qwencloud.com/developer-guides/clients-and-developer-tools/qoder)). Qoder's native selector is separate: availability and rate come from the account selector, not from that tier list ([Qoder](https://docs.qoder.com/qoder/model-selector)).

Qoder's selector lists `Qwen4-Max-Preview` at 0.5×. Actual Credits vary with the task, repository, parameters, and tool calls, so read the rate beside the model before sending ([Qoder](https://docs.qoder.com/qoder/model-selector)). The BYOK option in Qoder (custom models) is a connection mechanism, not a separate Qwen4 access product. It can point Qoder at a compatible Token Plan endpoint from the plan provider; a generic provider key does not by itself add Qwen4 access ([Qoder custom models](https://docs.qoder.com/qoder/custom-models)).

Choose native Qoder for the in-editor selector and Qoder Credits. Choose a Token Plan when you need a plan key, API access, and a compatible client with a custom Base URL. The two bill separately even with the same model ID.

For the family timeline before this preview, see the [Qwen 4 release date article](https://omniakey.com/blog/qwen-4-release-date).

## Beijing or Singapore?

Model Studio's Qwen4 page gives the Beijing host `token-plan.cn-beijing.maas.aliyuncs.com` and calls the plans Token Plan Personal and Token Plan Team ([Model Studio](https://help.aliyun.com/en/model-studio/qwen4-max-preview)). QwenCloud's latest-model page gives `token-plan.maas.qwencloudapi.com` ([QwenCloud](https://docs.qwencloud.com/developer-guides/getting-started/latest-model)). QwenCloud's Token Plan overview names the plans Token Plan Individual and Team Edition ([Token Plan](https://docs.qwencloud.com/token-plan/overview)).

QwenCloud sells Token Plan only in Singapore ([QwenCloud Token Plan](https://docs.qwencloud.com/token-plan/overview)). Its plan keys start with `sk-sp-`, while pay-as-you-go keys start with `sk-` or `sk-ws-`; keep each key on its own host ([QwenCloud Token Plan](https://docs.qwencloud.com/token-plan/overview), [Regions and endpoints](https://docs.qwencloud.com/developer-guides/getting-started/regions)).

For the complete protocol endpoints, the [QwenCloud regions page](https://docs.qwencloud.com/developer-guides/getting-started/regions) lists the OpenAI-compatible and Anthropic-compatible URLs for `token-plan.maas.qwencloudapi.com`. Model Studio's Qwen4 page points to its Token Plan guide for the protocol path.

## When does the preview end?

Model Studio says the Token Plan preview runs from 10:00 on November 11, 2026 to 24:00 on November 25, 2026 (UTC+8). The model is retired after the preview ends, and Alibaba says no changes will be made during the window ([English](https://help.aliyun.com/en/model-studio/qwen4-max-preview); [Chinese](https://help.aliyun.com/zh/model-studio/qwen4-max-preview)). QwenCloud's latest-model page lists the same dates ([QwenCloud](https://docs.qwencloud.com/developer-guides/getting-started/latest-model)).

Qoder's selector lists Qwen4-Max-Preview but does not show a preview-expiry label ([Qoder](https://docs.qoder.com/qoder/model-selector)). Whether the Qoder listing ends on November 25 is unstated on that page. Check the selector in your account before you plan work past that date.

Treat November 25 as the end of the Token Plan preview and check the model list afterward for what Alibaba offers next. The checked pages do not name a successor, so keep a fallback ready.

Status checked November 11, 2026.

## What limits matter for coding and agents?

Alibaba lists a 1M-token context window and a 393,216-token maximum output, shared by thinking and the final answer ([Model Studio](https://help.aliyun.com/en/model-studio/qwen4-max-preview)). Tokenize your input to see whether it fits the 1M window; a long repository can still exceed it. Thinking also uses the same output budget as the answer.

`max_tokens` is equivalent to `max_completion_tokens` from this model onward: both cap thinking plus answer, and earlier Qwen models keep their old behavior ([Model Studio](https://help.aliyun.com/en/model-studio/qwen4-max-preview)). Set the limit to match the reasoning and answer the task needs, not simply to the maximum.

Thinking is on by default. `reasoning_effort` takes low, medium, or xhigh, with xhigh as the default. Do not set `reasoning_effort` and `thinking_budget` together ([Model Studio](https://help.aliyun.com/en/model-studio/qwen4-max-preview)).

In multi-turn Chat Completions, `preserve_thinking` is on by default. Return historical thinking in the assistant `reasoning_content` field and keep it out of `content` ([Model Studio](https://help.aliyun.com/en/model-studio/qwen4-max-preview)).

The model supports function calling, web search, built-in tools, JSON Object and JSON Schema output, plus OpenAI Chat Completions, Responses, and Anthropic Messages APIs ([Model Studio](https://help.aliyun.com/en/model-studio/qwen4-max-preview)). Alibaba's recommended-models page tags it for complex reasoning and coding under Token Plan ([recommended models](https://www.alibabacloud.com/help/en/model-studio/models)).

The parameter notes above come from Alibaba's documentation; no model call was made.

## What should you check before a long coding task?

Match the access choice to the job:

- Native Qoder: check the selector and the displayed rate. The multiplier is not a total.
- Token Plan API: use the plan's host and key; for Qwen4-specific client steps, follow the Token Plan docs for the chosen region.
- Coding tools: the [best LLM coding agents](https://omniakey.com/blog/best-llm-coding-agents) comparison covers the wider field.

For general custom API configuration background, see the [Claude Code setup guide](https://omniakey.com/blog/claude-code-setup). It covers custom endpoints, not Qwen4-specific steps; use the official Token Plan docs for those.

## What can you use through OmniaKey today?

OmniaKey's catalog lists [Qwen3.8-Max](https://omniakey.com/qwen3-8-max); Qwen4-Max-Preview is absent. For a current Qwen coding model through OmniaKey, use Qwen3.8-Max on the OpenAI-compatible API.

The preview's Token Plan options are separate purchases from an OmniaKey key. If you want one key across several models, start with Qwen3.8-Max and add other models as your work needs them. See the [Qwen3.8-Max model page](https://omniakey.com/qwen3-8-max) for current access and details.

## Frequently asked questions

### Is Qwen4-Max-Preview available on the pay-as-you-go API?

No. Alibaba's Model Studio page says it is available through Token Plan only and that pay-as-you-go access is not supported ([Model Studio](https://help.aliyun.com/en/model-studio/qwen4-max-preview)).

### How long does the Qwen4-Max-Preview preview last?

The Token Plan preview runs from 10:00 on November 11, 2026 to 24:00 on November 25, 2026 (UTC+8), and the model is retired afterward ([Model Studio](https://help.aliyun.com/en/model-studio/qwen4-max-preview)). Qoder's native listing does not show an expiry date.

### Can I use Qwen4-Max-Preview in Qoder without a Token Plan?

Yes, if your Qoder account selector lists it; native use bills Qoder Credits at the displayed rate ([Qoder](https://docs.qoder.com/qoder/model-selector)). A QwenCloud Token Plan connection is a separate option and requires an eligible Qoder tier ([QwenCloud Qoder guide](https://docs.qwencloud.com/developer-guides/clients-and-developer-tools/qoder)).

### What are the context and output limits?

The context window is 1M tokens, and maximum output is 393,216 tokens shared by thinking and the final answer ([Model Studio](https://help.aliyun.com/en/model-studio/qwen4-max-preview)).

## Sources checked

- [Alibaba Cloud Model Studio: qwen4-max-preview](https://help.aliyun.com/en/model-studio/qwen4-max-preview)
- [阿里云百炼：qwen4-max-preview](https://help.aliyun.com/zh/model-studio/qwen4-max-preview)
- [Alibaba Cloud Model Studio: recommended models](https://www.alibabacloud.com/help/en/model-studio/models)
- [Qoder: Model Selector](https://docs.qoder.com/qoder/model-selector)
- [QwenCloud: Latest model: Qwen4-Max-Preview](https://docs.qwencloud.com/developer-guides/getting-started/latest-model)
- [QwenCloud: Token Plan](https://docs.qwencloud.com/token-plan/overview)
- [Qoder: Custom models](https://docs.qoder.com/qoder/custom-models)
- [QwenCloud: Regions, deployment scopes, and endpoints](https://docs.qwencloud.com/developer-guides/getting-started/regions)
- [QwenCloud: Qoder integration](https://docs.qwencloud.com/developer-guides/clients-and-developer-tools/qoder)