Alibaba Opens Preview Access to Qwen3.8-Max Through Qwen Cloud Token Plan

Alibaba has made Qwen3.8-Max-Preview available through Qwen Cloud’s Token Plan, according to the company’s official product page and documentation.

Qwen3.8-Max-Preview appears in official Qwen Cloud materials

Qwen Cloud’s official site advertises access to advanced models, including Qwen3.8-Max-Preview, through its Token Plan. The company’s Token Plan documentation lists qwen3.8-max-preview as supported in both Personal and Team editions and describes it as a preview model.

The documentation also says usage is subject to limited-time discounted pricing during the preview period. Qwen Cloud’s product page separately says most models offer free API credits to help users start without an upfront commitment, although the public materials reviewed do not establish that all Qwen3.8-Max-Preview usage is unlimited or permanently free.

Reports point to a 2.4-trillion-parameter flagship model

SiliconANGLE reported that Alibaba previewed Qwen3.8-Max-Preview at WAIC in Shanghai and described it as a 2.4-trillion-parameter flagship model available through Token Plan, Qoder, and QoderWork. TechNode also reported that Alibaba’s Qwen team outlined Qwen3.8 with 2.4 trillion parameters and said Qwen3.8-Max-Preview is available through Alibaba’s Token Plan, Qoder, and QoderWork.

Those reports position Qwen3.8 as part of Alibaba’s broader push to expand access to its Qwen model family across cloud, development, and work-assistant products. The official Qwen Cloud documentation is more narrowly focused on how customers can access the preview model through Token Plan editions.

What users can verify today

Based on the official Qwen Cloud pages, users can confirm that Qwen3.8-Max-Preview is listed as a Token Plan model and that Qwen Cloud is promoting starter access with free API credits for most models. The Token Plan documentation is the clearest source for model availability and preview status.

However, several claims circulating around the launch require caution. The official sources cited here do not provide public benchmark tables for Qwen3.8-Max-Preview, nor do they substantiate detailed performance comparisons against other frontier models. SiliconANGLE reported Alibaba’s claim that the model is second only to Claude Fable 5, but independent evaluations were not provided in the cited materials.

Why it matters

Preview access gives developers a way to test Alibaba’s latest Qwen model before broader production usage patterns are established. For teams already using Qwen Cloud, the Token Plan listing provides a straightforward path to experiment with the preview model under the terms described in Alibaba’s documentation.

The practical significance will depend on factors that are not yet fully documented in public sources, including real-world latency, reliability, pricing after the preview period, context limits, tool support, and independent benchmark results. Until Alibaba publishes more technical detail or third-party evaluations become available, Qwen3.8-Max-Preview is best understood as a newly available preview model rather than a fully characterized production benchmark leader.