IT Home reported on September 6 that last night, Alibaba quietly launched the Qwen3-Max-Preview model on the Tongyi Qianwen official website and OpenRouter, claiming it to be the most powerful language model in the Tongyi Qianwen series. Subsequently, the Tongyi large model introduced the features of this model through its official WeChat account.
The newly released Qwen3-Max-Preview (Instruct) has achieved significant enhancements compared to the 2.5 series in terms of Chinese and English understanding, complex instruction adherence, and tool invocation, while also greatly reducing knowledge hallucinations, making the model smarter and more reliable.
According to official sources, its parameter count reaches 1T, and it is leading the rankings in the Arena-Hard v2 benchmark test that measures the ability to tackle complex challenges. It also scored 80.6 in the AIME25 benchmark test, showcasing its strong logical reasoning. This model is expected to bring a whole new experience in handling complex workflows and conducting high-quality open-ended conversations.
IT Home provides the following experience links:
Qwen Chat: https://chat.qwen.ai/ Alibaba Cloud Bailian API Service: https://bailian.console.aliyun.com/(search for Qwen3-Max-Preview)
The official announcement from Alibaba Cloud Bailian today indicates that this update adopts a tiered billing model based on input length.返回搜狐，查看更多
平台声明：该文观点仅代表作者本人，搜狐号系信息发布平台，搜狐仅提供信息存储空间服务。