Recently, Alibaba launched its largest language model to date - Qwen3-Max-Preview. This model has over one trillion parameters, setting a new benchmark in AI technology. Qwen3-Max-Preview is now available to users through Qwen Chat and the Alibaba Cloud API. According to Alibaba, the model outperforms the previous top model Qwen3-235B-A22B-2507 in internal benchmark tests and early user feedback.
Qwen3-Max-Preview shows significant improvements in multiple aspects, including knowledge understanding, conversation ability, task processing, and instruction execution. At the same time, the model's performance in "knowledge hallucination" has been effectively reduced, meaning the information it provides is more accurate and reliable.
The model supports processing up to 258,048 input tokens and generating up to 32,768 output tokens. This high-capacity processing capability enables Qwen3-Max-Preview to perform exceptionally well in complex conversations and task execution. For users with needs, the pricing strategy is $2,151 per million input tokens, and $8,602 per million output tokens. It should be noted that Qwen3-Max-Preview currently does not support image processing functions.
With these powerful features, Alibaba hopes to further promote the development of the natural language processing field and provide users with more intelligent AI solutions. This innovative language model will bring more possibilities for enterprises and developers in application scenarios, helping them succeed on the path of digital transformation.
Experience entrance: https://chat.qwen.ai/
API: https://modelstudio.console.alibabacloud.com/?tab=doc#/doc/?type=model&url=2840914_2&modelId=qwen3-max-preview
Key points:
🌟 Qwen3-Max-Preview is Alibaba's latest released language model, with over one trillion parameters.
💬 The model significantly outperforms the previous generation model in aspects such as knowledge understanding, conversation ability, and task processing.
💰 The pricing is $2,151 per million input tokens, and $8,602 per million output tokens. It currently does not support image processing.
