An internecine AI battle has erupted between Alibaba and DeepSeek. Days after DeepSeek dominated several news cycles with its affordable DeepSeek-R1 reasoning model and the multimodal Janus-Pro-7B, Alibaba released its latest LLM, Qwen 2.5-Max, available via API from Alibaba Cloud. As with DeepSeek, Alibaba is looking beyond its domestic borders, but the fact that a public-facing AI battle is heating up between Chinese companies indicates the People’s Republic isn’t going to quietly cede the AI race to the U.S. Alibaba claims Qwen 2.5-Max outperforms models from DeepSeek, Meta and OpenAI.

In addition to the benchmarks “Qwen2.5-Max also demonstrates competitive results against industry leaders like OpenAI’s GPT-4o and Anthropic’s Claude-3.5-Sonnet in tests of advanced reasoning and knowledge,” VentureBeat reports, explaining that coming less than a week after the excitement over DeepSeek’s news, Qwen 2.5-Max had the effect of “further rattling U.S. technology markets and intensifying concerns about America’s eroding AI leadership.”

Qwen2.5-Max was announced in a blog post by Alibaba Cloud, which encompasses the AI division of Chinese e-commerce giant Alibaba Group Ltd.

“Continuously scaling both data size and model size can lead to significant improvements in model intelligence. However, the research and industry community has limited experience in effectively scaling extremely large models,” the blog post notes. “Many critical details regarding this scaling process were only disclosed with the recent release of DeepSeek-V3,” the general purpose LLM released in December.

Of its new Mixture-of-Experts (MoE) model, Alibaba Cloud says it has been pretrained on more than  20 trillion tokens “and further post-trained with curated Supervised Fine-Tuning (SFT) and Reinforcement Learning from Human Feedback (RLHF) methodologies.”

An MoE model “is an LLM architecture that uses multiple specialized models working in concert to handle complex tasks more efficiently according to a specific subset of expertise,” reports SiliconANGLE. “It’s essentially as if a team of AI models, each trained to excel in a specific subcategory of knowledge, work together to combine their training to answer questions and complete tasks.”

In addition to Claude and GPT-4o, Alibaba Cloud says the approach enabled the Qwen model to outperform DeepSeek-V3, as well as Meta Platforms’ Llama 3.1 401B across a range of benchmarks including MMLU-Pro (testing knowledge through college-level problems), LiveCodeBench (assessing coding ability), LiveBench (measuring general capabilities) and Arena Hard (approximating human preferences). Results are featured in the blog post.

Tom’s Guide suggests “DeepSeek’s emergence has been nothing short of disruptive and Alibaba’s new release is sure to cause a stir as well.”