Alibaba’s Qwen 2.5-Max Challenges GPT-4 & Llama 3.1, Boosting AI Race & Stock Gains

Chinese tech giant Alibaba (9988.HK) has introduced its latest artificial intelligence model, Qwen 2.5-Max, claiming it surpasses DeepSeek-V3, OpenAI’s GPT-4o, and Meta’s Llama-3.1-405B.

The timing of the release, coinciding with the first day of the Lunar New Year—a period when most Chinese citizens are off work—highlights the urgency Alibaba feels to compete against the rising influence of Chinese AI startup DeepSeek.

DeepSeek’s Meteoric Rise Disrupts the AI Landscape

The rapid ascent of DeepSeek has sent shockwaves through the AI industry.

Its AI assistant, launched on January 10, followed by the R1 model on January 20, has caused disruptions in Silicon Valley, leading to plunging tech stocks.

Investors are now questioning the massive expenditures by leading U.S.-based AI companies, given DeepSeek’s ability to produce competitive models at significantly lower costs.

DeepSeek’s success has also pressured domestic AI companies in China to accelerate their own model advancements.

ByteDance, the parent company of TikTok, reacted swiftly by updating its flagship AI model just two days after DeepSeek-R1’s debut.

ByteDance claims its updated model outperforms OpenAI’s o1 model in the AIME benchmark, a measure of AI comprehension and response accuracy.

The AI Model Price War in China

DeepSeek’s previous model, DeepSeek-V2, triggered a pricing war in China’s AI sector when it was introduced in May.

Priced at an unprecedented 1 yuan ($0.14) per million tokens, it forced Alibaba’s cloud unit to slash prices on its AI models by up to 97%.

Other major Chinese tech firms, including Baidu (9888.HK) and Tencent (0700.HK), also adjusted their pricing strategies to compete.

Liang Wenfeng, DeepSeek’s elusive founder, stated in a rare July interview with Chinese media outlet Waves that DeepSeek prioritizes achieving artificial general intelligence (AGI) over engaging in price wars.

He emphasized that traditional tech giants, with their high costs and rigid corporate structures, may struggle to keep pace with DeepSeek’s agile and research-driven approach.

Alibaba’s Qwen 2.5: Features and Capabilities

Alibaba’s Qwen 2.5 is a versatile AI model designed to handle text, images, and audio.

It integrates seamlessly into Qwen Chat, allowing users to interact with different versions such as Qwen 2.5 Plus, Qwen 2.5 Max, and Qwen2.5-VL-72B-Instruct.

Competing with OpenAI’s ChatGPT, Google Gemini, and Microsoft Copilot, Qwen 2.5 boasts:

• Multilingual support: Covers 29 languages, including English, Chinese, Spanish, French, Japanese, Korean, and Arabic.
• Enhanced context processing: Handles up to 128K tokens in a single conversation, improving comprehension of complex topics.
• Structured data processing: Works with tables and JSON formats, enhancing data analysis and interpretation.
• Multimodal capabilities: Processes both images and audio, enabling diverse applications such as creative content generation and advanced data insights.
• AI agent functions: Supports chatbot role-play, interactive behaviors, and condition-based responses, making it adaptable for various scenarios.

DeepSeek’s Challenges Despite Rapid Growth

Despite its rapid rise, DeepSeek has faced criticism for its model’s performance.

A recent audit by NewsGuard ranked DeepSeek’s chatbot tenth out of eleven AI models in terms of accuracy. The study found that DeepSeek’s chatbot had:

• 17% accuracy in delivering news and information.
• 30% false claims repetition rate.
• 53% vague or unhelpful responses.
• An 83% fail rate overall.

These findings raise concerns about DeepSeek’s readiness to compete at the highest levels against more established AI firms like OpenAI and Google.

The Future of AI Competition

With Alibaba’s Qwen 2.5 entering the race, China’s AI industry is poised for fierce competition.

While DeepSeek’s low-cost, high-performance models have shaken up the market, Alibaba’s extensive resources and robust cloud infrastructure may give it an edge in the long run.

Meanwhile, ByteDance, Tencent, and Baidu continue to refine their AI offerings, ensuring that China remains a global AI powerhouse.

The coming months will reveal whether DeepSeek can sustain its disruptive momentum or if Alibaba and other tech giants will reclaim dominance.

Regardless, the battle for AI supremacy in China—and globally—has never been more intense.