Investing.com-- Alibaba (NYSE:BABA) rolled out a new artificial intelligence model on Wednesday- Qwen2.5-Max- and claimed that it outperformed major rivals such as OpenAI’s GPT-4o, DeepSeek-V3, and Meta’s Lama-3.1.
The company's shares in Hong Kong rose 1.2%, while the US-listed stock added 1% in premarket trading Wednesday.
Qwen2.5-Max comes amid increased focus on Chinese AI models, after the release of DeepSeek-V3 and DeepSeek R1, earlier in January, sparked increased questions over U.S. dominance in artificial intelligence.
Alibaba said Qwen2.5-Max was launched on the Alibaba Cloud Bailian platform, and that developers could use a new chat platform to interact directly with the model.
It claimed that the new model outperformed DeepSeek-V3 on several major benchmarks, including Arena-Hard, LiveBench, LiveCodeBench and MMLU-Pro.
Alibaba said Qwen2.5-Max is comparable to Anthropic's flagship Claude-3.5-Sonnet, and “almost comprehensively surpasses GPT-4o, DeepSeek-V3, and Llama-3.1-405B.”
Chinese AI models have come into renewed focus in recent weeks, especially after DeepSeek’s R1 model sparked questions over just how much the billions of dollars being poured into AI infrastructure was justified.
DeepSeek appears to match offerings from its American rivals while using older hardware and a much lower budget. The model’s release wiped out about $1 trillion in U.S. tech stock valuations on Monday, with Nvidia (NASDAQ:NVDA) accounting for a bulk of losses.
The prospect of companies adopting leaner and more efficient AI models sparked concerns that AI-fueled demand for chips will not be as strong as initially expected.
Such a scenario also draws into question the billions of dollars invested in AI infrastructure by Wall Street’s so-called AI hyperscalers- primarily Microsoft Corporation (NASDAQ:MSFT) and META (NASDAQ:META).
Comments