Home » Alibaba Unveils Flagship Qwen4-Max Model With 2.4 Trillion Parameters

Alibaba Unveils Flagship Qwen4-Max Model With 2.4 Trillion Parameters

By

–

Alibaba Unveils Flagship Qwen4-Max Model With 2.4 Trillion Parameters

Alibaba’s Qwen4-Max model has been making waves in the AI community with its 2.4 trillion parameters and lower cloud API pricing, but what does it take to run this model locally? The Qwen team at Alibaba announced the top-tier model, Qwen4-Max, on November 16th, Japan Standard Time, featuring a new MoE architecture, the ability to read images, and a context window of 1 million tokens.

Background And History

The Qwen series, also known as Tongyi Qianwen, is a family of predominantly open-weights large and small language models developed by Alibaba Cloud, with its latest version being Qwen3.8. The Qwen series has been downloaded more than 40 million times, with its models having been used to develop fine-tuned versions. The Qwen4-Max model is the newest and most capable version of the Qwen family of AI models, featuring 2.4 trillion parameters, native vision capabilities, and a one-million-token context window, as reported by PYMNTS.com.

How It Works

Qwen4-Max can be used via the QwenCloud API, Qwen Studio, QwenWork, and Qoder, with pricing set at $1.50 per 1 million tokens for input and $4.50 for output, according to the official announcement. The model also writes safer code, with its pass rate on the Last Secure Code Bench for code that both works and is secure rising from 22 percent to 31.8 percent, as stated by Digital Trends. To run Qwen4-Max locally, developers can use the QwenCloud API or Token Plan, and the model is also available in Qwen Work, Qoder, and Qwen Studio.

Quantization Requirements

While the exact quantization requirements for running Qwen4-Max locally are not specified, the model is designed to be used as a cloud service, and smaller open-source Qwen4 models are expected to arrive in the future. The Qwen-VL series, a line of visual language models, has been released with variants of 2 billion and 7 billion parameters, and the Qwen2.5-VL was released with variants of 3, 7, 32, and 72 billion parameters, as reported by Wikipedia. The Qwen4-Max model itself has 2.4 trillion parameters but activates only 71 billion per token, as stated by Digital Trends.

Hardware Requirements

Running Qwen4-Max locally requires significant hardware resources, although the exact requirements are not specified. However, the Qwen2.5-VL model, which has variants of 3, 7, 32, and 72 billion parameters, can be run on devices with sufficient memory and processing power. The Qwen-VL series combines a vision transformer with an LLM, and the Qwen2.5-VL model has been released with variants of 3, 7, 32, and 72 billion parameters, as reported by Wikipedia.

According to Digital Trends, the Qwen4-Max model can handle the same jobs as other top-of-the-line AI models for much less money, with the average cost per task on the in-house Work360 benchmark dropping from $4.43 to $1.67 compared to Qwen3.8-Max. This represents a 62 percent reduction in task costs, making it an attractive option for developers and businesses looking to integrate AI into their applications.

Sources

Photo: Douglas Lopes / Unsplash (Unsplash License)

Get breaking stories first on our Telegram channel.

More stories by

Most read

AI Tools

Artificial Intelligence

Business

World News