appConfigEntityId

bodyType

customContents

entityId

entityUuid

isVideo

articlePageQuery

article

articleSchemaContent

content

hooksArticlePageAdvertisementConfigArticle

hooksArticlePageFloatingButtonConfigArticle

appConfig

filter

json

liveArticle

Xinmei Shen

Alibaba Group Holding’s new lightweight AI model Qwen3.8-27B has matched much larger near-frontier rivals while being able to run on everyday hardware, impressing developers as local AI gains momentum.
The Qwen3.8-27B, a small model with 27 billion parameters, performed on par with OpenAI’s GPT-5.6 Luna, which was billed as the most cost-efficient model in the US lab’s latest flagship series, benchmark firm Artificial Analysis said on Monday.
It nearly matched leading open-weight models from Chinese firms, including DeepSeek-V4-Pro-0813, released last week with 1.7 trillion parameters, and Zhipu’s 753-billion-parameter GLM-5.2 launched in June, according to the Artificial Analysis Intelligence Index.
OpenAI, which launched the GPT-5.6 family last month, does not disclose the parameter counts of its models.
The new findings came days after Alibaba released Qwen3.8-27B’s model weights – the underlying parameters that encode its intelligence – last Friday.
On Artificial Analysis’ Agentic Index, which measures models’ performance in AI agent-focused workflows, Alibaba’s small model outperformed GPT-5.6 series’ mid-tier model Terra and Anthropic’s powerful Claude Opus 4.8 released in May.
The large language model (LLM) represents “the first time a local model has scored frontier model capability”, the San Francisco-based open-source AI coding agent platform Cline said on social media on Tuesday.
The Qwen3.8-27B offered an “excellent size for running a model on a reasonably specced (specification) laptop”, said Simon Willison, British programmer and developer of Datasette, an open source tool for exploring and publishing data, in a blog post published on Sunday.
Willison added that he had been running Qwen3.8-27B on a 128-gigabyte M5 Max MacBook Pro and an Nvidia DGX Spark, and described it as a “miracle” that the model, sporting a modest 17GB file, could “generally do everything that I need from an LLM for getting real work done”.


“The models at this size continue to get better at an impressive rate,” Willison wrote. “We don’t need to spend half a million dollars on data centre-class hardware just to run a competent model.”
Still, the new Qwen model “defaults to wildly overthinking things” and appeared slow in performance, according to Willison.
Global developers’ interest in running AI locally on their own devices has been expanding as smaller open-weight models become increasingly powerful.
“We weren’t expecting this pace of local progress anywhere near this soon,” Cline said of Alibaba’s new lightweight model on Tuesday.
The Qwen3.8-27B was released last week alongside Qwen3.8-Max, Alibaba’s latest flagship model with a massive 2.4-trillion-parameter size and performance nearly matching the current open-weight leader Kimi K3.
Alibaba has so far open sourced more than 460 Qwen models with total global downloads surpassing 3 billion, the company said on social media last Friday.
Qwen saw more than 2 billion downloads in 2026, far ahead of all other AI model providers, according to a report published last Friday by open source developer platform Hugging Face.
Google’s open models saw 418 million downloads this year, while those from OpenAI and Meta Platforms had 329 million and 227 million downloads respectively, according to Hugging Face. DeepSeek models recorded 186 million downloads.
Qwen benefited from a consistent pace of flagship model release, as well as the fact that its models cover a wide range of sizes and use cases, which allows developers to “stay within the same ecosystem whether they need a small local model or a larger deployment model”, Hugging Face said.
Alibaba is the owner of the South China Morning Post.


Philippine tribe vows to stand ground against US-led Pax Silica: ‘listen to us’

As the US weighs restrictions on cloud computing, how will China’s AI sector adapt?

Chinese AI chipmaker Hygon plots expansion from data centres to robotics

For China’s hi-tech IPOs, it’s Hong Kong vs the mainland – but how much does it matter?

Artificial intelligence (AI)

Alibaba’s lightweight Qwen model takes on larger AI systems from OpenAI, DeepSeek, Zhipu

Chinese tech titan Alibaba Group Holding’s new lightweight AI model Qwen3.8-27B has matched much larger rivals including OpenAI, DeepSeek and Zhipu. Photo: Shutterstock

Zhipu AI’s answer to Project Glasswing marks shift for Chinese cyber safety: researcher

The logo of Chinese AI start-up Zhipu AI on a smartphone. Photo: CFOTO/Future Publishing via Getty Images

The permanent exhibition at the Zhongguancun Exhibition Centre in Beijing, which showcases over 560 technologies and products from more than 350 companies. Photo: Xinhua

The next silicon? AI data centre material faces price spike amid China supply crunch

Indium phosphide is used to make lasers that convert electrical signals into light, crucial to the development of fibre-optic cables. Photo: Shutterstock

Alibaba sells gaming arm Lingxi Games as tech giant sharpens focus on AI and e-commerce

Three Kingdoms Tactics, a war-based strategy game, is Lingxi’s most popular title. Photo: Lingxi Games

China’s AI start-up funding triples to US$16b in first quarter amid bets on LLMs, robotics

AI-related start-ups in China secured more than 110 billion yuan in the first three months of the year. Photo: Shutterstock Images

Moonshot AI founder Yang Zhilin delivers a speech at the Zhongguancun forum in Beijing, March 25, 2026. Photo: Reuters

Chinese Premier Li Qiang visits a hi-tech enterprise during an inspection tour in Beijing, May 18, 2026. Photo: Xinhua

Next level: why China’s game makers are quietly bankrolling generative AI

Peacekeeper Elite, also known as Game for Peace, by Tencent Holdings. Photo: Handout

MiHoYo has been using MiniMax’s LLMs in the development of its popular title Honkai: Star Rail since 2023. Photo: Shutterstock

Tech

Tech Trends

Big Tech

News

China Future Tech

Chinese tech giant’s latest offering performed on par with OpenAI’s GPT-5.6 Luna and nearly matched DeepSeek, Zhipu’s open-weight models


Chinese tech giant’s latest offering performed on par with OpenAI’s GPT-5.6 Luna and nearly matched DeepSeek, Zhipu’s open-weight models.


Alibaba’s lightweight Qwen model takes on larger AI systems from OpenAI, DeepSeek, Zhipu

.css-1c6uqr6{color:inherit;font-weight:inherit;font-size:inherit;font-family:inherit;line-height:inherit;overflow-wrap:break-word;}Chinese tech giant’s latest offering performed on par with OpenAI’s GPT-5.6 Luna and nearly matched DeepSeek, Zhipu’s open-weight models

Chinese tech giant’s latest offering performed on par with OpenAI’s GPT-5.6 Luna and nearly matched DeepSeek, Zhipu’s open-weight models