Hermes Vector

TechTrending

Alibaba Launches Qwen4-Max AI Model With Strict Cybersecurity Guardrails

Alibaba has launched Qwen4-Max, a closed AI model with 2.4 trillion parameters and strict cybersecurity guardrails, while planning Qwen 5 at 5 to 10 trillion parameters.

By Hermes Vector

Also in: Español · Português · Français · 中文

Checked against 2 sourcesscmp.comnews.lavx.hu
Illustration for: Alibaba Launches Qwen4-Max AI Model With Strict Cybersecurity Guardrails

In brief

  • Alibaba launched Qwen4-Max, a closed AI model with 2.4 trillion parameters and strict cybersecurity guardrails.
  • Qwen4-Max activates 71 billion parameters per token, 25 per cent less than its predecessor, cutting average task cost by 60 per cent.
  • Alibaba plans to scale its next model, Qwen 5, to between 5 trillion and 10 trillion parameters.
  • Tencent launched Hy4 Flash, a 224 billion parameter model priced at 7 US cents per million input tokens.

Alibaba Launches Qwen4-Max AI Model With Strict Cybersecurity Guardrails

Alibaba launches Qwen4-Max

Alibaba Group Holding has launched its most advanced artificial intelligence model to date, Qwen4-Max, with strict cybersecurity guardrails, as the tech giant balances safety with a continued commitment to open sourcing amid growing global concerns over AI-enabled cyberattacks.

The release highlights a growing focus on safety in China’s AI industry as Alibaba said it intentionally filtered the model’s training data to remove offensive cyberattack capabilities while retaining capabilities useful for cyber defence such as vulnerability detection.

The Hangzhou-based company also introduced dedicated data sets and post-training enhancements to mitigate chemical, biological, radiological and nuclear (CBRN) risks, it said on Monday.

Alibaba also highlighted the model’s improvements in scientific research, including the completion of a continuous 48-hour run to discover an experimentally validated coolant molecule for next-generation semiconductor chips, as well as autonomous research on Alzheimer’s disease.

There is heated debate about whether open-weight models, which are favoured by Chinese companies, could heighten CBRN risks more than the closed-source models championed by their US rivals such as OpenAI and Anthropic.

Some experts have warned that AI systems could eventually be powerful enough to enable malicious actors or states to develop CBRN weapons.

While Qwen4-Max is a closed model, Alibaba reaffirmed its open-source strategy by noting that smaller-scale, open-weight versions of the Qwen4 series would be released to global developers soon.

Alibaba’s Max series of models had always been closed-source until Qwen 3.8, which was made available as open source in August. The Max series is also the company’s largest model by parameter size. However, the company highlighted its achievements in making its latest model more cost-efficient when served to users.

Qwen4-Max boasts 2.4 trillion parameters, the same size as its previous Qwen3.8-Max, though it activates only 71 billion parameters per token – 25 per cent less than its predecessor.

That allows it to deliver superior reasoning capabilities while bringing down inference costs, cutting the average cost per task by 60 per cent, according to Alibaba, citing an internal benchmark called Work360, which evaluates model performance on various work tasks.

The model is currently available via QwenCloud, with input pricing set at about 8 US cents per 1 million tokens.

Alibaba’s AI efforts have focused heavily on capturing the enterprise AI market and cultivating an open-source community to outpace local rivals like DeepSeek and Z.ai, as well as US competitors.

Liu Dayiheng, head of Alibaba’s Qwen large language model team, presented a road map outlining plans for the Qwen 4, Qwen 4.5 and Qwen 5 series in September, while stressing the team’s commitment to the open-source community earlier this month at the World Internet Conference in Wuzhen.

The company has said that its next-generation model, Qwen 5, would be scaled up to between 5 trillion and 10 trillion parameters.

Alibaba owns the South China Morning Post.

Tencent launches Hy4 Flash

Separately, Shenzhen-based Tencent Holdings rolled out Hy4 Flash on Monday, a lightweight and low-cost model with 224 billion parameters priced at 7 US cents per million input tokens.

Described as the most cost-effective Hy model to date, the flash variant comes less than two weeks after the launch of Tencent’s latest flagship model Hy4, which the company ranks among the world’s first-tier open-weights models.

Tencent said the flash model only cost one-twentieth as much per task as the flagship Hy4, and one-eighth as much as DeepSeek-V4.1-Flash-1017 from Hangzhou-based rival DeepSeek.


Sources

This article was drafted with AI assistance and checked against the sources above. Company claims are reported as claims. Cover image is AI-generated.

Questions readers ask

What is Qwen4-Max and what cybersecurity measures does it include?
Qwen4-Max is Alibaba Group Holding's most advanced artificial intelligence model, launched with strict cybersecurity guardrails that intentionally filter training data to remove offensive cyberattack capabilities while retaining useful cyber defense functions.
How does Qwen4-Max compare to its predecessor in terms of parameters and cost?
Qwen4-Max boasts 2.4 trillion parameters, the same size as Qwen3.8-Max, but activates only 71 billion parameters per token, which cuts the average cost per task by 60 per cent.
What are the pricing details for accessing Qwen4-Max?
The model is available via QwenCloud with input pricing set at about 8 US cents per 1 million tokens.
What future models has Alibaba announced for the Qwen series?
Alibaba plans to release smaller-scale, open-weight versions of the Qwen4 series soon and has stated that its next-generation model, Qwen 5, would be scaled up to between 5 trillion and 10 trillion parameters.
What is Tencent's Hy4 Flash and how does its cost compare to competitors?
Tencent Holdings rolled out Hy4 Flash, a lightweight model with 224 billion parameters priced at 7 US cents per million input tokens, which costs one-twentieth as much per task as the flagship Hy4 and one-eighth as much as DeepSeek-V4.1-Flash-1017.

The day's checked news, by email

One short email a day with the top stories. Free, and you can unsubscribe in one click.

Or follow for the latest news:

How do you feel about this story?

Comments 0

No comments yet. Start the discussion.