Overview
Qwen3.8-Max is Alibaba’s flagship multimodal foundation model, announced on August 3, 2026. It uses a sparse mixture-of-experts architecture with 2.4 trillion total parameters and a hybrid attention mechanism. Alibaba made it available through Model Studio APIs and QwenWork. (alibabacloud.com)
Capabilities and uses
The model accepts text, images, and video and produces text. It supports thinking and non-thinking modes, function calling, structured output, web search, prefix completion, and context caching. Alibaba positions it for coding, office work, research, visual understanding, long-document and video analysis, and long-horizon planning and execution, including professional tasks in law, finance, and design. (help.aliyun.com)
Limits and versioning
- The documented context window is 1,000,000 tokens. Maximum input is 991,808 tokens, or 983,616 in thinking mode; maximum output is 131,072 tokens. Supported lengths can vary with API parameters. (docs.modelstudio.console.alibabacloud.com)
- Model tuning is not supported. Batch availability varies by region and model version. Outputs should be checked, particularly for professional or consequential decisions. (help.aliyun.com)
- Alibaba announced that the
qwen3.8-maxendpoint would transition to snapshotqwen3.8-max-0902on September 5, 2026, with pricing unchanged. A snapshot can be specified separately when version stability matters. (alibabacloud.com)
API pricing
The following are official Model Studio list prices, checked November 16, 2026. Prices are per 1 million tokens, with input and output billed separately. They apply to qwen3.8-max in thinking and non-thinking modes for the documented 0–1M input-token tier; output billing includes thinking tokens and answers. Domestic CNY and international USD prices are separate regional offerings, not currency conversions. (help.aliyun.com)
| Official platform and region | Standard input | Standard output |
|---|---|---|
| Mainland China site, Beijing | CNY 12 | CNY 36 |
| International site, Singapore, International deployment | USD 2 | USD 6 |
| International site, US Virginia, US deployment | USD 2 | USD 6 |
| International site, global deployment in Hong Kong, Frankfurt, Virginia, or Tokyo | USD 1.65 | USD 4.951 |
For the mainland Beijing offering, the listed Batch File rates are CNY 6 input and CNY 18 output per million tokens; implicit cache-hit input is CNY 1.5 per million. Batch Chat is listed at the standard CNY 12 input and CNY 36 output rates. Conditions and availability differ by operation and region. (help.aliyun.com)
Official pricing sources: Mainland China model pricing; international model pricing. Checked November 16, 2026.