appConfigEntityId

bodyType

customContents

entityId

entityUuid

isVideo

articlePageQuery

article

articleSchemaContent

content

hooksArticlePageAdvertisementConfigArticle

hooksArticlePageFloatingButtonConfigArticle

appConfig

filter

json

liveArticle

Vincent Chow

Ben Jiang

Tencent Holdings has released a new flagship artificial intelligence model, the first since former OpenAI researcher Yao Shunyu joined the Chinese tech giant to lead its foundational AI development efforts.
The Shenzhen-based company said on Thursday that the new model, an open-source program called “Hy3 preview”, was its most powerful yet, on par with top Chinese models but still lagging flagship products from US leaders such as OpenAI and Google DeepMind.
Notably, the model is relatively small with only 295 billion parameters, bucking a recent trend of large models with trillions of parameters – the mathematical variables that encode a model’s “intelligence”, which are roughly proportional to the amount of computational power needed to train and serve the model.
The company’s previous flagship model, the HY 2.0 released in early December, had over 400 billion parameters.
The new model was developed to cater for real-world business scenarios, the company said, highlighting the collaboration between Tencent’s foundational model development team Hunyuan and its Yuanbao AI application team.
“By seamlessly aligning product-side requirements with underlying technology, we have successfully bridged the gap between model capability and user value,” it said, adding that the model was already deployed in Tencent’s flagship AI products including consumer app Yuanbao and coding assistant CodeBuddy.
Agentic capabilities were “one of the most significantly improved areas in this release”, with the company pointing to strong performance on several in-house benchmarks it had developed to assess compatibility with popular agentic AI tool OpenClaw.
Tencent has been one of the most enthusiastic Chinese tech giants to embrace OpenClaw. In March, the company held a high-profile event to help members of the public install the tool on their own devices.

Since then, it has further integrated OpenClaw offerings into its cloud platform and introduced an international version of its consumer app QClaw on Tuesday, which allows users to customise OpenClaw-like AI agents of their own.
Tencent launched its flagship Hunyuan model series in 2023, months after OpenAI released ChatGPT, which kicked off the generative AI boom.
However, in the past year, Hunyuan’s overall capabilities have been surpassed by domestic rivals including Zhipu AI’s GLM and Moonshot AI’s Kimi, triggering an internal restructuring that saw the arrival of Yao late last year to take up the mantle of chief AI scientist.
According to Thursday’s announcement, Tencent’s latest model marked a “reconstruction” of the Hunyuan model series, with the model taking less than three months to develop.
During this period, Tencent said it rebuilt its foundational model development process, including both pre-training and post-training, the two main stages of AI model training.
The company said it “aims to gather comprehensive user feedback” through the release of this preview version to “further enhance the practicality” of the full Hy3 model, which has no release date yet.
In his first public appearance as Tencent’s new AI lead in January, Yao suggested that Tencent would place its massive consumer user base at the centre of its AI strategy, with the company’s flagship super app WeChat boasting over 1 billion users.
Under this strategy, the importance of the foundational model lies not in the number of parameters it has but the amount of “context” it knows about its user, such as preferences and background, according to Yao.
“Imagine forwarding your WeChat chat history to an app like Yuanbao: suddenly, the model has much richer context to work with, and the value to the user increases significantly,” he said at the time.


Huawei quickens AI chip pace, promises next entrant 3 quarters early

Hong Kong’s ‘AI city brain’ will not tap personal data, breach privacy: John Lee

Huawei unveils latest tech to boost AI power in push to break China’s Nvidia reliance

Hong Kong’s 5-year plan, policy continuity offer investors more stability, experts say

Artificial intelligence (AI)

Tencent unveils first flagship AI model with former OpenAI researcher at helm

Tencent headquarters in Shenzhen. Photo: Shutterstock Images

Yao Shunyu was a high-profile recruit poached by Tencent from OpenAI earlier this year. Photo: Handout

Moonshot AI releases flagship model as open-source push continues

Founder of Moonshot AI, Yang Zhilin, delivers a speech the 2026 Zhongguancun forum in Beijing, China, March 25, 2026. Photo: Tingshu Wang

Differences in growth and maturity are shaping the commitment of Chinese AI companies to open source. Photo: Shutterstock

Can ‘token exports’ give China an edge in the AI era?

Starting this year, models from Chinese companies have intermittently ranked among the global top tier in token usage. Photo: SOPA Images/LightRocket via Getty Images

The OpenAI logo is seen in this illustration taken in 2024. Photo: Reuters

The Qwen booth at the Mobile World Congress (MWC) in Barcelona, Spain, March 2. Photo: Xinuua

Tech

Big Tech

News

China Future Tech

The model is relatively small with only 295 billion parameters, bucking a recent trend of large models with trillions of parameters


The model is relatively small with only 295 billion parameters, bucking a recent trend of large models with trillions of parameters.


Tencent unveils first flagship AI model with former OpenAI researcher at helm

.css-1c6uqr6{color:inherit;font-weight:inherit;font-size:inherit;font-family:inherit;line-height:inherit;overflow-wrap:break-word;}The model is relatively small with only 295 billion parameters, bucking a recent trend of large models with trillions of parameters

The model is relatively small with only 295 billion parameters, bucking a recent trend of large models with trillions of parameters