In a significant development within the artificial intelligence landscape, Chinese tech giant Tencent appears to have onboarded the WizardLM team, a group of AI researchers previously associated with Microsoft. This move is particularly noteworthy given WizardLM’s recent history and highlights the intense competition for top AI talent and technology. Tencent’s Strategic Move in AI Tencent, the powerhouse behind platforms like WeChat and popular games including PUBG Mobile, is making aggressive strides in artificial intelligence. The company’s AI development organization, Hunyuan, is reportedly the new home for the WizardLM researchers. This strategic acquisition underscores Tencent’s commitment to bolstering its AI capabilities and accelerating its model development efforts. Can Xu, a senior AI researcher who played a leading role in several WizardLM projects, announced the team’s departure from Microsoft to join Tencent’s Hunyuan group via a post on X. Understanding WizardLM’s Past at Microsoft AI The WizardLM team, which originated several years ago, has a somewhat unusual past, particularly during its time under the Microsoft AI umbrella. In April 2024, the team released the WizardLM-2 family of AI models, which they claimed were competitive with leading models like OpenAI’s GPT-4. However, these models were quickly pulled from public access by Microsoft after just one day. The reason cited was a missed step in the release process: toxicity testing. The team acknowledged this oversight on X, stating they were completing the testing and planned to re-release the models soon. The Fallout from WizardLM-2’s Removal Despite Microsoft’s swift action, users had already downloaded and reuploaded the original WizardLM-2 models, along with customized versions. The sudden removal also caused disruption within the open-source AI community. Hugging Face CEO Clément Delangue noted that the deletion of WizardLM models, which were downloaded hundreds of thousands of times monthly, harmed numerous open-source projects dependent on them. This incident highlighted the complexities and responsibilities involved in releasing powerful AI models, even for established tech giants like Microsoft. Tencent’s Growing Hunyuan AI Power With the integration of the WizardLM team, Tencent’s Hunyuan AI efforts are expected to gain further momentum. Hunyuan has recently launched various AI models, including generators for video and 3D objects. The collaboration has already yielded results, with the release of the Hunyuan-TurboS 0416 model. Qingfeng Sun, identifying as a “co-founder” of WizardLM, has claimed that Hunyuan-TurboS 0416 surpasses certain “open” AI models, specifically mentioning Google’s Gemma 3 series, in performance. This suggests that the WizardLM team is continuing its focus on developing high-performing AI models within its new environment. Tencent has significantly increased its investment in AI infrastructure and recently restructured the Hunyuan team into two new units. The company attributed an 8% year-over-year growth in Q1 2025 partly to its AI investments. Tencent plans substantial capital expenditures this year, approximately 90 billion yuan (around $12.49 billion), with a significant portion allocated to fuel its AI ambitions. The addition of the WizardLM team aligns with this aggressive investment strategy aimed at strengthening Tencent’s position in the global AI race. Implications for the Future of AI Models The acquisition of a skilled team like WizardLM by Tencent signifies the ongoing talent war in the AI sector. While the exact number of researchers moving to Tencent is unclear, the integration of their expertise into the Hunyuan platform is likely to accelerate the development of new AI models. The competition between major players like Tencent, Microsoft, Google, and others continues to drive innovation. The rapid release and comparison of models like Hunyuan-TurboS 0416 against competitors underscore the fast-paced nature of this field. The move also raises questions about the future direction of open-source contributions from teams previously associated with large corporations and the balance between rapid development and responsible deployment, as highlighted by the WizardLM-2 incident. In conclusion, Tencent’s reported hiring of the WizardLM team from Microsoft is a major development that strengthens its Hunyuan AI division. Despite WizardLM’s controversial history surrounding the release and subsequent withdrawal of its models at Microsoft, the team’s expertise is now set to contribute to Tencent’s ambitious AI goals. This move highlights the intense competition for AI talent and resources, promising further advancements and shifts in the landscape of AI model development. To learn more about the latest AI models and market trends, explore our articles on key developments shaping the AI landscape.
