Microsoft AI Team Move: Tencent Hires WizardLM Research Group

Microsoft AI Team Move: Tencent Hires WizardLM Research Group

Microsoft AI Team Move: Tencent Hires WizardLM Research Group

Tencent has acquired the WizardLM team, a Beijing-based Microsoft AI research group known for its advanced language models and a turbulent history. The move, confirmed by senior AI researcher Can Xu, signals Tencent’s aggressive push into AI development under its Hunyuan division. WizardLM has already released a Hunyuan-branded model, Hunyuan-TurboS 0416, which reportedly outperforms open models like Google’s Gemma 3 series.

The transition follows WizardLM’s abrupt removal of its WizardLM-2 models from public access last year due to missing toxicity testing-a misstep that drew criticism from the open-source community. Now under Tencent, the team is expected to continue developing cutting-edge AI models, backed by the company’s $12.49 billion capital expenditure plan for AI infrastructure.

Key Implications of the Microsoft AI Team Move

The WizardLM team’s shift from Microsoft to Tencent underscores the intensifying competition for AI talent in China. Tencent’s Hunyuan division, which recently underwent restructuring, is positioning itself as a major player in generative AI, with models spanning text, video, and 3D object generation. The hiring of WizardLM-a group with prior experience in high-performance AI systems-could accelerate Tencent’s ambitions to rival global leaders like OpenAI and Google DeepMind.

Qingfeng Sun, a co-founder of WizardLM, claims that Hunyuan-TurboS 0416 surpasses open-source alternatives, though independent benchmarks are yet to verify this. Meanwhile, Tencent’s Q1 2025 financial report highlights an 8% revenue growth attributed to AI investments, reinforcing the strategic importance of this Microsoft AI team move.

WizardLM’s Controversial Past

WizardLM’s tenure at Microsoft was marked by both innovation and controversy. The team’s WizardLM-2 models, briefly released in April 2024, were pulled within a day after Microsoft cited incomplete toxicity testing. Despite the takedown, users quickly reuploaded the models, leading to disruptions in open-source projects reliant on them. Hugging Face CEO Clément Delangue criticized the decision, noting that the models had been downloaded over 100,000 times monthly.

Now at Tencent, WizardLM faces fewer constraints, but scrutiny over AI safety and performance will likely persist. The team’s ability to deliver reliable, high-performing models under Hunyuan’s banner will be closely watched.

Frequently Asked Questions

Why did WizardLM leave Microsoft?

The exact reasons remain undisclosed, but the team’s move aligns with Tencent’s heavy investment in AI and its restructuring of the Hunyuan division.

What is Hunyuan-TurboS 0416?

It’s an AI model developed by WizardLM under Tencent’s Hunyuan brand, claimed to outperform open models like Google’s Gemma 3 series.

How does this affect Microsoft’s AI strategy?

Microsoft retains other AI research teams, but losing WizardLM may slow progress in certain specialized language model developments.