Home NATIONAL NEWS Alibaba launches Qwen4-Max, claims it outperforms US AI rivals in coding and...

Alibaba launches Qwen4-Max, claims it outperforms US AI rivals in coding and cybersecurity

4
0

Source : INDIA TODAY NEWS

Alibaba Group has launched Qwen4-Max, its latest flagship artificial intelligence (AI) model, as it continues to rival US AI players such as Anthropic and OpenAI. The Chinese tech giant claims its new model brings significant improvements in coding, cybersecurity, professional work and scientific research. It is designed to handle complex tasks, from writing and debugging code to building games, analysing financial documents and assisting with scientific experiments.

advertisement

Unlike some of Alibaba’s earlier releases, Qwen4-Max is a closed-source model, meaning developers cannot download its underlying weights and run it independently. However, the company says smaller models in the Qwen4 family will be released as open source.

What is new with Qwen4-Max?

Qwen4-Max is built on Alibaba’s new Qwen4 architecture and has a 2.4-trillion-parameter backbone, with 71 billion parameters activated for each token processed. In simple terms, it uses a Mixture-of-Experts approach, where only a portion of the model is activated for a particular task rather than using the entire model every time. This, according to the company, helps the model to handle demanding workloads while keeping computing requirements lower.

The model also supports a one-million-token context window, allowing it to process large amounts of information in a single conversation. It comes with native vision capabilities, which means it can understand visual information alongside text.

Alibaba says Qwen4-Max is more efficient than its predecessor, Qwen3.8-Max, using up to 35 per cent fewer tokens on its Work360 benchmark and 49 per cent fewer on Terminal-Bench 4.0. The company also reports a 62 per cent reduction in average spending per task on Work360, from $4.43 to $1.67.

Qwen4-Max vs GPT-6.1-Sol and Anthropic Opus 5.5

As always Alibaba has also compared Qwen4-Max with the leading models including OpenAI’s GPT-6.1-Sol and GPT-6 Astra, Anthropic’s Opus 5.5 and Kimi K3.1 across several benchmarks.

According to the company’s internal testing in coding, Qwen4-Max scored 75.7 on DeepSWE v1.1, compared with 74.2 for Opus 5.5 and 75.2 for GPT-6.1-Sol.

Alongside its capabilities in coding, professional tasks, creative work and scientific research, Alibaba says Qwen4-Max is also better equipped to handle cybersecurity tasks. It can identify software weaknesses and suggest ways to fix them, while blocking information that could help people launch cyberattacks. The company has also added safeguards to prevent the model from being misused in areas such as chemical, biological and nuclear threats.

Qwen4-Max availability and pricing

Qwen4-Max is available through Alibaba’s QwenCloud, QwenWork, Qoder and QwenStudio platforms. The company has also introduced Live Delivery, a feature that allows the model to deliver an early working version of a project and improve it through subsequent updates based on user feedback. Alibaba is open-sourcing the associated Qwen-Live-Delivery-Plugin.

The model’s cached-input API pricing starts at 0.5 yuan, approximately $0.075 (around Rs 7), per million tokens. Alibaba says the Qwen4 family will also include smaller open-source models, although the company has not detailed their complete release schedule.

– Ends

SOURCE :- TIMES OF INDIA