eFinder

eFinder

Home › AI Technology Development › Article

Tencent unveils first flagship AI model under tenure of former OpenAI researcher

South China Morning Post · · 204 words · By Vincent Chow
AI Technology Development
open_in_new Read the original article: https://www.scmp.com/tech/big-tech/article/3351101/tencent-unveils-first-flagshi…

psychologyDetected Techniques

warning
Loaded Language 60% confidence
Using words with strong emotional connotations to influence an audience.

fact_checkFact-Check Results

7 claims extracted and verified against multiple sources including cross-references, web search, and Wikipedia.

info Single Source 5
check_circle Corroborated 2
info
“Tencent unveils first flagship AI model with former OpenAI researcher at helm”
SINGLE SOURCE
The web search results mention Tencent's restructuring and the involvement of a former OpenAI researcher in charge of AI R&D, but none of the provided evidence explicitly confirms that Tencent 'unveiled its first flagship AI model' with this researcher 'at the helm' in a single, verifiable announcement. The evidence points to the *personnel change* and the *effort*, but not the specific unveiling described.
menu_book
wikipedia NEUTRAL — Kling AI is a generative artificial intelligence service created and hosted by the Beijing-based technology company Kuaishou. Kling generates videos from natural language descriptions, called prompts.…
https://en.wikipedia.org/wiki/Kling_AI
menu_book
wikipedia NEUTRAL — Moonshot AI (Moonshot; Chinese: 月之暗面; pinyin: Yuè Zhī Ànmiàn; lit. 'Dark Side of the Moon') is an artificial intelligence (AI) company based in Beijing, China. It has been dubbed one of China's "AI Ti…
https://en.wikipedia.org/wiki/Moonshot_AI
menu_book
wikipedia NEUTRAL — Knowledge Atlas Technology Joint Stock Co., Ltd., branded internationally as Z.ai, is a Chinese technology company specializing in artificial intelligence (AI). The company was formerly known as Zhipu…
https://en.wikipedia.org/wiki/Z.ai
+ 3 more evidence sources
info
“The model is relatively small with only 295 billion parameters, bucking a recent trend of large models with trillions of parameters”
SINGLE SOURCE
Two web search results mention parameter counts (389 billion parameters for Hunyuan-Large), but the claim specifies '295 billion parameters.' The evidence provides a different, specific number (389 billion) for the model mentioned in the context of the web search results, making the exact number in the claim unverified by the provided evidence.
menu_book
wikipedia NEUTRAL — Honor of Kings (simplified Chinese: 王者荣耀; traditional Chinese: 王者榮耀; pinyin: Wángzhě Róngyào, unofficially translated as King's Glory, or alternatively transliterated as Wangzhe Rongyao) is a multipla…
https://en.wikipedia.org/wiki/Honor_of_Kings
menu_book
wikipedia NEUTRAL — Shutterstock, Inc. is an American provider of stock photography, stock footage, stock music, and editing tools; it is headquartered in New York. Founded in 2002 by programmer and photographer Jon Orin…
https://en.wikipedia.org/wiki/Shutterstock
menu_book
wikipedia NEUTRAL — A virtual assistant (VA) is a software agent that can perform a range of tasks or services for a user based on user input, such as commands or questions, including verbal ones. Such technologies often…
https://en.wikipedia.org/wiki/Virtual_assistant
+ 3 more evidence sources
info
“The Shenzhen-based company said on Thursday that the closed-source HY3-Preview was its most powerful model yet, on par with leading Chinese models, while still lagging top US models from the likes of OpenAI and Google DeepMind.”
SINGLE SOURCE
The web search results mention 'Top Close-source Model' and 'Top Close-source Model3' related to Hunyuan3D, suggesting a high-end, closed-source model. However, none of the provided evidence explicitly states that the model is 'HY3-Preview,' nor does it provide direct comparisons stating it 'lags top US models from the likes of OpenAI and Google DeepMind' in the manner claimed. This specific comparison and naming convention are not corroborated.
travel_explore
web search NEUTRAL — Top Close-source Model2.We designed a diffusers-like API to use our shape generation model - Hunyuan3D-DiT and texture synthesis model - Hunyuan3D-Paint. You could assess Hunyuan3D-DiT via: from hy3dg…
https://huggingface.co/tencent/Hunyuan3D-2
travel_explore
web search NEUTRAL — Alibaba's Qwen3.6-Plus is now available on API Yi — MoE architecture with 72B params (18B effective), 1M token context, Terminal-Bench 61.6 surpassing Claude Opus 4.5, top-tier coding Agent capabiliti…
https://docs.apiyi.com/en/news/qwen-3-6-plus-launch
travel_explore
web search NEUTRAL — Top Close-source Model3. 3.218.We designed a diffusers-like API to use our shape generation model - Hunyuan3D-DiT and texture synthesis model - Hunyuan3D-Paint. You could assess Hunyuan3D-DiT via: fro…
https://github.com/Tencent-Hunyuan/Hunyuan3D-2
check_circle
“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.”
CORROBORATED
The web search results provide general information on LLMs and parameters. One source states that training and deploying models with billions of parameters requires significant computational resources, which supports the proportionality concept. This general principle is supported by the context provided in the search results regarding computational cost.
travel_explore
web search NEUTRAL — A large language model (LLM) is a computational model designed to perform natural language processing tasks, especially language generation, using contextual relationships derived from a large set of …
https://en.wikipedia.org/wiki/Large_language_model
travel_explore
web search NEUTRAL — Computational Cost: Training and deploying models with billions of parameters requires significant computational resources. It may take days or weeks to train these models on large datasets, requiring…
https://www.geeksforgeeks.org/artificial-intelligence/what-a…
travel_explore
web search NEUTRAL — The Types of Parameters that Power AI Models. Parameters can be broken down into three main categories: weights, biases, and embeddings. Each serves a unique purpose within the model: Weights: These r…
https://aiverge.ai/parameters-in-large-language-models-the-s…
info
“The company’s previous flagship model, the HY 2.0 released in early December, had over 400 billion parameters.”
SINGLE SOURCE
The evidence mentions Tencent's previous flagship model, HY2.0, and that it was released, but it does not specify that it had 'over 400 billion parameters.' The evidence is insufficient to confirm the exact parameter count stated in the claim.
menu_book
wikipedia NEUTRAL — The 70th anniversary of the founding of the People's Republic of China (Chinese: 庆祝中华人民共和国成立70周年) was observed with a series of ceremonial events including a grand military parade as its spotlight to …
https://en.wikipedia.org/wiki/70th_anniversary_of_the_People…
menu_book
wikipedia NEUTRAL — Wang Xinyu (born 26 September 2001) is a Chinese professional tennis player. Wang reached a career-high singles ranking of world No. 30 on 23 February 2026, and set a doubles ranking of No. 16 on 20 M…
https://en.wikipedia.org/wiki/Wang_Xinyu
menu_book
wikipedia NEUTRAL — The Xi'an Y-20 Kunpeng (Chinese: 运-20 鲲鹏; pinyin: Yùn-20 Kūnpéng; lit. 'Transport-20 Kunpeng') is a large military transport aircraft that was developed by the Xi'an Aircraft Industrial Corporation fo…
https://en.wikipedia.org/wiki/Xi'an_Y-20
+ 3 more evidence sources
info
“HY3-Preview 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.”
SINGLE SOURCE
The web search results confirm that Tencent's model development involves Hunyuan and that Yuanbao is a key product. One result mentions DeepSeek-R1 supporting both HunYuan and DeepSeek models within Yuanbao, indicating integration for business scenarios. However, the evidence does not explicitly detail the formal 'collaboration' structure between a dedicated 'Hunyuan foundational model development team' and a 'Yuanbao AI application team' as a development mandate.
travel_explore
web search NEUTRAL — Tencent's self-developed large model Hunyuan 2.0 (Tencent HY2.0) has been officially released. At the same time, DeepSeek V3.2 is gradually integrated into Tencent's ecosystem.
https://www.aibase.com/news/1189
travel_explore
web search NEUTRAL — Business insights.misc{hunyuan3d22025tencent, title={Hunyuan3D 2.0: Scaling Diffusion Models for High Resolution Textured 3D Assets Generation}, author={Tencent Hunyuan3D Team}, year={2025}, eprint={2…
https://github.com/Tencent-Hunyuan/Hunyuan3D-2
travel_explore
web search NEUTRAL — Since then, Tencent Yuanbao has welcomed a series of rapid iterations and feature updates: On February 13, it integrated the full version of DeepSeek-R1, supporting both HunYuan and DeepSeek models
https://longportapp.com/en/news/229268878
check_circle
“the model was already deployed in Tencent’s flagship AI products including consumer app Yuanbao and coding assistant CodeBuddy.”
CORROBORATED
Multiple web search results confirm that the model (or related products) are integrated into Tencent's AI ecosystem, specifically naming Yuanbao and CodeBuddy. This is reported across different contexts within the search results.
travel_explore
web search NEUTRAL — Tencent Cloud CodeBuddy is a next-generation AI Code Editor. Powered by the Tencent Yuanbao Code Large Model, it provides services such as code completion, error diagnosis, technical Q&A, and performa…
https://www.codebuddy.ai/
travel_explore
web search NEUTRAL — On April 24th, Tencent Cloud announced a significant upgrade to its code assistant, CodeBuddy, launching the Craft intelligent software development agent. This innovative tool elevates AI programming …
https://www.aibase.com/news/17478
travel_explore
web search NEUTRAL — At the conference, Tencent debuted the 'Tencent Cloud Intelligent Agent Strategic Panorama', the Tencent Cloud Intelligent Agent Development Platform ADP3.0, and the Hunyuan 3D 3.0 base model; it also…
https://www.sohu.com/a/935496888_122328931
info Disclaimer: This analysis is generated by AI and should be used as a starting point for critical thinking, not as definitive truth. Claims are verified against publicly available sources. Always consult the original article and additional sources for complete context.