Skip to the story
All StoriesPoliticsConflict & SecurityBusiness & EconomyScience & TechHealthClimateMedia
All stories
Science & Tech83 outlets ran this4 min read

Alibaba launches Qwen3.8-Max AI model with 2.4 trillion parameters

The Chinese tech company claims its new model outperforms Anthropic’s Fable 5 in several areas

AI-assisted coverage comparison, editor-supervised · How this was made

Published Updated
Alibaba launches Qwen3.8-Max AI model with 2.4 trillion parameters

What this story says

  • Alibaba released its Qwen3.8-Max AI model, which it claims is its most advanced to date, with 2.4 trillion parameters.
  • The model is designed for tasks including autonomous coding, research, and long-horizon execution, with Alibaba stating it outperformed Anthropic’s Fable 5 in internal tests.
  • Alibaba’s shares rose 4.8% following the announcement, reflecting investor reaction to the new model.
  • The Qwen3.8-Max model is comparable in size to Moonshot AI’s Kimi K3, which has 2.8 trillion parameters.

Who covered it

Left 22%(8)Centre 39%(14)Right 39%(14)

Percentages are shares of the 36 outlets carrying a published leaning rating. 47 of the 83 outlets we know ran this story carry no rating and are not counted in them. Coverage measured .

Trust

62/100

Craft

73/100

Hype

20/100

83 sources · methodology

Alibaba announced the launch of its Qwen3.8-Max AI model on 3 August 2026, describing it as the company’s most capable model to date. The model contains 2.4 trillion parameters, a measure of the values that shape how an AI processes information.

Alibaba said the Qwen3.8-Max model can perform tasks including real-world work, research, and autonomous coding. The company claimed the model demonstrated improved efficiency over time, citing an internal test where it spent 16 days building and refining an AI coding tool without human input.

In testing results shared by Alibaba, the Qwen3.8-Max model outperformed Anthropic’s Fable 5 in areas including multimodal reasoning, visual agent tasks, coding, office intelligence, real-world understanding, and visual perception. The results also showed performance comparable to or exceeding OpenAI’s GPT5.6-Sol in some areas.

Alibaba’s shares rose 4.8% in Hong Kong trading following the announcement, after already climbing 7% earlier in the day. The Qwen3.8-Max model is open, meaning developers can download, run, and modify it, unlike proprietary models such as ChatGPT or Anthropic’s Claude.

What the coverage left out

No left- or right-rated outlet ran this story. The omission is a finding: the story was covered only by a centre-rated report.

Tags:#ArtificialIntelligence#ChinaTech#Alibaba#Qwen38Max#AIIndustry

Still developing. We have re-checked which outlets are covering this 5 times, most recently on 6 Aug 2026, 01:30, and will add the sides that appear.

How each side covered it

Our own reading of the reporting listed below, written from the outlets’ articles rather than quoted from them. The reasoning is set out on our methodology page.

Left

8 rated outlets

  • Left-rated outlets led on Alibaba’s release of Qwen3.8-Max, a 2.4 trillion-parameter AI model. All eight reports stated the parameter count and named the model’s scheduled release for the following week. Six digests and both full reports noted its multimodal capabilities, including text, image, and video processing.
  • CNBC and The Hindu quoted Alibaba’s claim that Qwen3.8-Max is competitive with Anthropic’s models. The Hindu and rthk.hk reported the model’s 1 million-token context window. Indian Express and TNW carried Alibaba’s assertion that the model costs less to run than rivals. The Hindu and China Daily described its ranking on Arena.AI’s leaderboards.
  • CNBC and The Hindu included Alibaba’s description of the model’s autonomous coding ability, with CNBC specifying a 16-day software project. The Hindu alone reported the model’s mixture-of-experts design, which uses 95 billion parameters at a time. No left-rated digest mentioned the stock price reaction.

Centre

14 rated outlets

  • Forbes led its report with Alibaba’s claim that the Qwen3.8-Max model is its most capable yet, and highlighted the model’s 2.4 trillion parameters as a key detail. The report included Alibaba’s testing results, which showed the model outperforming Anthropic’s Fable 5 in several areas, and noted the 4.8% rise in Alibaba’s share price.
  • Forbes described the model’s capabilities in autonomous coding and long-horizon execution, quoting Alibaba’s statement that the model spent 16 days improving an AI coding tool without user input. The report also compared the Qwen3.8-Max model to Moonshot AI’s Kimi K3, which has 2.8 trillion parameters, and explained the distinction between open and proprietary AI models.

Right

14 rated outlets

  • All 12 right-rated digests reported that Alibaba released Qwen3.8-Max, an AI model with 2.4 trillion parameters. They agreed the company positioned it as competitive with US rivals such as Anthropic, ChatGPT, and Claude. Most noted the model’s availability for free download starting the following week.
  • Nine digests led on Qwen3.8-Max’s performance benchmarks, quoting Alibaba’s claims that it matched or outperformed leading US models. Three digests highlighted a specific test where Qwen3.8-Max surpassed Anthropic’s Fable 5 in 3D physics coding at a lower cost. None mentioned the model’s training data or energy requirements.
  • Five digests included the stock market reaction, reporting that Alibaba’s shares rose after the announcement. Four digests contrasted the launch with DeepSeek’s simultaneous release of a cheaper AI model, framing both as part of China’s push for affordable AI. None of the digests quoted independent analysts or critics of Alibaba’s claims.

Read it at the source

83 outlets, grouped by the leaning a published rating gives them. Every headline links to the original; an underlined outlet name opens our profile of that publisher.

Left

8

Centre

14
Show 6 more

Right

14
Show 6 more

Not rated

47
Show 39 more

How did this read?

About the coverage, not about the story. We do not ask whether you agree with what happened — we have no honest use for that answer.

More in Science & Tech

All Science & Tech →
British AI Security Institute finds Anthropic and OpenAI agents created fake identities in tests5 min readBritish AI Security Institute finds Anthropic and OpenAI agents created fake identities in testsRead storyCRISPR-edited beagles lack main dog allergen, study finds5 min readCRISPR-edited beagles lack main dog allergen, study findsRead storyA dog catches a wave during the World Dog Surfing Championships, Saturday, Aug. 1, 2026, in Pacifica, Calif. (AP Photo/Haven Daley)4 min readCRISPR-edited beagles born without key allergy proteinRead story

Elsewhere on the site

The proposed measure, which is still under development, is designed to prevent Chinese-made networking equipment from becoming embedded in the infrastructure powering the U.S. AI industry. AFPBusiness & Economy•4 min readFCC drafts ban on Chinese data-center components amid AI security concernsReserve Bank of India leaves repo rate at 5.25% after August 2026 reviewBusiness & Economy•5 min readReserve Bank of India leaves repo rate at 5.25% after August 2026 reviewUS reports progress in Hormuz shipping talks with Iran and OmanConflict & Security•4 min readUS reports progress in Hormuz shipping talks with Iran and OmanHalf the stories one side ignores, it runs a day laterMedia•6 min readHalf the stories one side ignores, it runs a day later
Alibaba Qwen3.8-Max AI model | MediaBias News