Alibaba Cloud launched its latest AI model, Qwen 2.5, on January 29, 2025. The company claims it definitively outperforms DeepSeek-V3 across multiple critical benchmarks.
The Chinese tech giant directly challenged the rising AI startup by demonstrating superior performance in key evaluation metrics. Alibaba explicitly stated that Qwen 2.5-Max surpasses DeepSeek-V3 in Arena-Hard, LiveBench, LiveCodeBench, and GPQA-Diamond tests.
The company’s announcement highlighted its model’s dominance over not just DeepSeek, but also other major AI models like OpenAI‘s GPT-4o and Meta’s Llama 3.1-405B.
The new model series spans from 0.5 to 72 billion parameters, offering unprecedented performance across different scales. Qwen 2.5 can handle up to 128,000 tokens and generate 8,000 tokens, supporting over 29 languages.
By making most Qwen 2.5 models open-source under the Apache 2.0 license, Alibaba aims to accelerate AI innovation. The strategic release directly challenges DeepSeek’s recent market disruption and positions Alibaba as a formidable competitor in the global AI landscape.
The tech industry is watching closely as Alibaba’s direct challenge to DeepSeek unfolds. The coming months will reveal the long-term implications of this bold performance claim and its potential to reshape the AI development ecosystem.
