Tencent rolls out Hy3 AI model globally via WorkBuddy, TokenHub
Synopsis
Tencent's Hy3 — a 295-billion-parameter model with 68x API growth over its predecessor — topped OpenRouter's global LLM leaderboard in its first week, and is now free worldwide on WorkBuddy until 31 August, marking the company's sharpest international AI push yet.
Key Takeaways
Tencent Holdings expanded global access to its flagship Hy3 AI model on 5 August 2026 via WorkBuddy , Tencent Design Miora , and TokenHub .
Hy3 is available free of charge on WorkBuddy for users worldwide until 31 August 2026 , according to Tencent .
The model carries 295 billion parameters , making it Tencent 's most powerful to date.
API call volume for Hy3 rose more than 68-fold versus the previous generation, and it topped the OpenRouter global LLM usage leaderboard in its first week.
The rollout is the first major AI strategy update since former OpenAI researcher Yao Shunyu joined Tencent to lead foundational AI development.
Poshu Yeung , Senior Vice-President of Tencent Cloud , framed Hy3 as a tool for enterprises to operationalise AI at scale.
Tencent Holdings has expanded international access to its flagship Hy3 large language model, making the 295-billion-parameter system available to global developers, enterprises, and users through three platforms on Wednesday, 5 August 2026. The rollout marks the Shenzhen-based technology giant's most assertive overseas AI push to date, positioning Hy3 directly against the world's leading frontier models.
What Tencent launched and where
Hy3 is now accessible through WorkBuddy, Tencent's workplace agentic tool; Tencent Design Miora, its creation studio; and TokenHub, its model platform. On WorkBuddy, the model will be available free of charge to users worldwide until 31 August, according to the company. The wider rollout follows the model's official launch last month.
Why it matters
Poshu Yeung, Senior Vice-President of Tencent Cloud and head of the unit's international business, said Hy3 would enable organisations to “move beyond AI experimentation and put intelligence to work in practical, trusted and scalable ways.” The deployment signals that Tencent is no longer content to lead only in its domestic market, joining rivals including Alibaba Group Holding, ByteDance, and Zhipu AI in targeting global enterprise customers.
Performance benchmarks and industry signals
Despite its relatively compact footprint, Hy3 is Tencent's most capable model to date. According to the company, API call volume for Hy3 surged more than 68-fold compared with the previous generation, and the model topped the global large language model usage leaderboard on OpenRouter in the first week of its launch. The release is being closely tracked by the industry as the first significant strategic AI update since former OpenAI researcher Yao Shunyu joined Tencent to lead its foundational AI development efforts.
The competitive backdrop
China's AI sector is increasingly outward-facing, with labs racing to capture developer mindshare on global platforms such as OpenRouter. DeepSeek's earlier international breakthrough demonstrated that Chinese models can compete on capability and cost, and Tencent's free-access window on WorkBuddy appears designed to replicate that adoption playbook. The move puts pressure on incumbents and signals a broader commoditisation of frontier-level AI access.
What's next
The free WorkBuddy access window closes on 31 August 2026, after which pricing and tier structures will come into focus. Analysts and enterprise buyers will be watching whether Tencent's international distribution partnerships deepen, and how Yao Shunyu's influence shapes the next model generation. The pace of API adoption on TokenHub will be a key indicator of whether Hy3 can sustain its early leaderboard position against fast-moving competitors.
Point of View
Build API dependency, then monetise. What mainstream coverage underplays is the significance of Yao Shunyu's role: a former OpenAI researcher now steering Tencent's foundational models is a direct talent-war dividend that could accelerate capability gains. The 68x API volume surge and the OpenRouter leaderboard position suggest Hy3 is not a vanity launch — it is a credible distribution bet. The deeper question is whether Western enterprise procurement teams, already navigating data-sovereignty concerns, will embrace a Shenzhen-origin model at scale.
NationPress
5 Aug 2026
Frequently Asked Questions
What is Tencent Hy3 and when was it launched globally?
Tencent Hy3 is a 295-billion-parameter large language model and the company's most powerful AI model to date. It was rolled out to global developers, enterprises, and users on 5 August 2026 through WorkBuddy , Tencent Design Miora , and TokenHub , following its official launch the previous month.
Is Tencent Hy3 free to use?
Hy3 is available free of charge on WorkBuddy for users worldwide until 31 August 2026 , according to Tencent . After that date, pricing terms have not yet been announced publicly.
How does Hy3 compare to other AI models?
Hy3 topped the global large language model usage leaderboard on OpenRouter in its first week of launch, and its API call volume increased more than 68-fold compared with the previous generation, according to the company. It is positioned to compete with frontier models from OpenAI , Alibaba Group Holding , and ByteDance .
Who is leading Tencent's AI development?
Former OpenAI researcher Yao Shunyu joined Tencent to lead its foundational AI development efforts. The Hy3 global rollout is regarded as the first major AI strategy update under his direction.
Why is Tencent pushing Hy3 internationally?
Tencent is joining a broader wave of Chinese AI companies — including DeepSeek , Alibaba Group Holding , ByteDance , and Zhipu AI — targeting global enterprise customers. Poshu Yeung , Senior Vice-President of Tencent Cloud , said Hy3 is designed to help organisations move beyond AI experimentation and deploy intelligence at scale.
