Elon Musk Teases Grok 3.5 After Alibaba Launches Qwen3 AI Model
- tech360.tv
- 11 hours ago
- 2 min read
Elon Musk revealed a new version of xAI’s Grok chatbot just hours after Alibaba Group Holding released its latest Qwen3 artificial intelligence models, intensifying the US-China AI rivalry.

On Tuesday, Alibaba launched the third generation of its Qwen AI family, including multiple versions with varying parameter sizes. The largest model, with 235 billion parameters, reportedly outperformed DeepSeek-R1 and OpenAI’s o1 reasoning models. A smaller 600 million-parameter version may be efficient enough to run on smartphones, according to some experts.
Shortly after Qwen3 was made available on the open-source platform Hugging Face, Musk announced that xAI would release an early beta of Grok 3.5 to SuperGrok subscribers. He claimed it is the first AI capable of accurately answering technical questions about rocket engines and electrochemistry.
The release follows a surge in foundational AI model development, sparked by the January debut of DeepSeek-R1. That model’s low cost and high performance prompted concerns in the US about losing its AI lead.
In the past three months, Chinese tech giants Baidu, ByteDance and Tencent Holdings have also updated their foundational models. These now rival or match the performance of US models such as Google’s Gemini 2.5 Pro, OpenAI’s o3 and o4, and Meta Platforms’ Llama 4.
A recent Stanford University report found that China is rapidly closing the gap with the US in developing advanced AI models. Alibaba’s Qwen has become the world’s largest open-source AI ecosystem, with over 100,000 derivative models, surpassing Meta’s Llama.
Qwen3 was released just three months after Qwen2.5-Max, underscoring the rapid pace of AI development. Meanwhile, speculation continues around DeepSeek’s upcoming R2 reasoning model.
The Qwen team, part of Alibaba Cloud, described the new model as a major step toward achieving artificial general intelligence, where AI matches or exceeds human intelligence.
Nathan Lambert, a researcher at the US-based Allen Institute for AI, said open-source Chinese models like Qwen3 and DeepSeek-R1 could be key to gaining market share in the US. He noted that these companies are effectively influencing the American AI ecosystem through technological innovation.
Elon Musk announced Grok 3.5 hours after Alibaba launched Qwen3
Qwen3 includes a 235 billion-parameter model that outperforms rivals
Grok 3.5 can answer technical questions on rocket engines and electrochemistry
Source: SCMP
Comments