Muse Ember
Meta’s multimodal model handles repeated coding subtasks, tool calls, routing and document processing inside agent workflows.
Qwen4-Max Preview
www.alibabacloud.com
A time-limited Qwen reasoning model for coding, visual understanding and interactive agent tasks through Alibaba Cloud Token Plan.
Visit website
Pricing
Paid
Available on
Alibaba Cloud Token Plan, Interactive coding tools, Agent clients
Category
Llm Models
What is Qwen4-Max Preview?
Qwen4-Max Preview is Alibaba Cloud's preview model for complex reasoning, coding and visual understanding. International access is provided through Token Plan for developers evaluating assistance in supported interactive coding and agent clients.
Qwen Code offers CLI, desktop and IDE configuration routes, including a Singapore Token Plan configuration for this model. The preview ends November 25, 2026 at 24:00 UTC+8, after which Alibaba says the model will be retired. Its short availability window makes it an evaluation option with a defined end date.
Key features
Reasoning and coding assistance
Alibaba's international model catalogue lists the preview for complex reasoning and coding tasks through Token Plan.
Visual understanding
The international model catalogue identifies visual understanding as a supported capability, allowing developers to evaluate visual context alongside coding tasks.
Qwen Code client options
Qwen Code documents CLI, desktop and IDE setup routes, including qwen4-max-preview on the Singapore Token Plan endpoint.
Who is it for?
Developers investigating code
Discuss an implementation problem interactively and assess proposed changes.
Frontend developers discussing visual context
Evaluate whether visual understanding helps explain an interface or implementation question.
Teams evaluating an interactive assistant
Compare the model's usefulness for reasoning and coding tasks within a supported client during the preview.
How to use Qwen4-Max Preview
- Review Token Plan eligibility, permitted usage and the preview end date before subscribing.
- In Qwen Code, select Alibaba ModelStudio, choose Token Plan and select Singapore.
- Configure your dedicated Token Plan API key in the client.
- Select qwen4-max-preview and begin an interactive coding or agent conversation.
Qwen4-Max Preview pricing
Paid
Token Plan Personal is a subscription with monthly Credits. Consumption varies with model use, tokens, thinking and tool calls. Unused monthly Credits do not roll over, and service pauses when the quota and any purchased extra Credits are exhausted.
Check current pricing ↗What to know before you use it
- The preview will be retired after November 25, 2026 at 24:00 UTC+8. Access requires Token Plan; pay-as-you-go access is unavailable.
- Token Plan Personal permits interactive coding and agent use only. Automation scripts, custom application backends and non-interactive batch calls are prohibited.
- The international plan is offered in Singapore with global inference and cross-border transfer of prompts and outputs.
- Qwen Code defaults to a 32,768-token output limit for this model; longer output requires configuration. Alibaba's mainland model specification describes a shared budget for thinking and the final answer. International parity for that detailed budget behavior is not confirmed.
Frequently asked questions
Can I use this preview as an application backend?
Token Plan Personal prohibits custom application backends and non-interactive batch calls.
Which model identifier should I select?
The international Qwen Code configuration lists qwen4-max-preview for the Singapore Token Plan endpoint.
Does Singapore access keep inference in Singapore?
No. The plan uses global inference and involves cross-border data transfer.