China’s Alibaba and Baidu step up global competition with new reasoning-focused AI models

Alibaba and Baidu are intensifying the battle for leadership in China’s fast-evolving AI market, with both companies unveiling upgraded models focused on dynamic reasoning capabilities.

On Tuesday, Alibaba launched Qwen 3, an enhanced version of its flagship AI model. The upgrade introduces hybrid reasoning, designed to improve adaptability and efficiency for app and software developers.

It follows a rapid-fire development pace from Alibaba, which pushed out Qwen 2.5-Max in January, just days after startup DeepSeek shook the industry by showcasing high-performing models at lower costs compared to Western competitors.

Meanwhile, Baidu, the Chinese search giant, introduced two new models late last week: Ernie 4.5 Turbo and the reasoning-focused Ernie X1 Turbo. Both models aim to enhance performance in complex decision-making and multi-step problem-solving — areas increasingly seen as critical for enterprise adoption of AI technologies.

Comparable to Western rivals

The back-to-back launches highlight growing competition in China’s AI sector, as domestic tech firms race not only against each other but also to keep pace with Western rivals such as OpenAI, Anthropic, and Google DeepMind.

Alibaba’s Qwen 3 includes multiple models, notably its 235-billion-parameter flagship Qwen3-235B-A22B and a smaller 30-billion-parameter Mixture of Experts version, Qwen3-30B-A3B. Both models will be released with open weights.

“In initial benchmarks, both of these models have been measured as being roughly on par with OpenAI and DeepSeek models and slightly behind Grok 3 beta and Google Gemini 2.5 Pro,” said Hyoun Park, CEO and chief analyst at Amalgam Insights. “Baidu’s Ernie 4.5 Turbo Measures up on par with OpenAI’s newest GPT models while being priced much more competitively than OpenAI.”

Analysts also point out that Chinese AI models are reaching performance levels comparable to Western counterparts at a fraction of the cost, between 20 and 40 times lower, putting pressure on US firms to speed up innovation and cut prices.

“However, ongoing geopolitical tensions are likely to restrict the use of Chinese models in regulated sectors,” said Prabhu Ram, VP of the industry research group at Cybermedia Research. “Consequently, established market players will need to respond to these emerging start-ups by increasing investment in domestic AI development, while managing higher operational costs in a significantly more fragmented and geopolitically complex technology landscape.”

The announcements also signal a broader shift in capabilities, highlighting advances beyond text-based models into multi-modal AI, according to Sharath Srinivasamurthy, associate vice president of Research at IDC.

“Another observation is that they are going all out to woo the developer community,” Srinivasamurthy said. “China has the largest developer community in the world, and greater mindshare among developers will lead to wider adoption of the technology. From a price and performance perspective, the race has started. We are already seeing the new announcements emphasize being better and cheaper, and the trend is expected to continue.”

‘Dynamic reasoning’ models for enterprise use cases

Alibaba’s latest release combines conventional AI capabilities with advanced dynamic reasoning, creating what the company describes as a more adaptable and efficient platform for app and software developers.

Dynamic and hybrid reasoning has quickly become one of the hottest trends in AI model development over the past few months, as companies seek to build systems capable of more complex and flexible problem-solving.

“Dynamic reasoning is allowing models to break down problems step-by-step and support more complex decision-making,” Ram said. “Emerging models such as Qwen 3 and Ernie X1 Turbo illustrate this transition, offering enterprises real-time adaptability, greater automation, and over 30% cost savings through innovations such as Mixture-of-Experts architectures and tool autonomy.” At the same time, Ram warned that as AI reasoning grows more dynamic, enterprises will face new challenges related to operational complexity, model reliability, and data governance, particularly when using models developed outside established regulatory frameworks.