Chinese search engine giant Baidu has launched its latest AI model, Ernie 4.5 Turbo, and a new reasoning model called Ernie X1 Turbo.

The announcement was made on April 25, 2025 as the company seeks to enhance its AI product offerings.

This launch occurs amid increasing competition in China’s AI market.

Various firms are striving to develop advanced technologies.

🔗 Source: Reuters

🧠 Food for thought

1️⃣ Baidu’s long-term AI vision shows strategic continuity amid tech evolution

The launch of Ernie 4.5 Turbo represents the latest step in Baidu’s decade-long AI journey, not a sudden pivot.

Baidu’s first ERNIE model was introduced in 2019, when it notably surpassed Google and Microsoft by scoring over 90 points on the GLUE benchmark for language understanding1.

This early success stemmed from ERNIE’s innovative approach to masking strings of characters rather than individual words, making it particularly effective for Chinese language processing while still performing well in English1.

The company’s AI foundation dates back even further, with significant speech recognition breakthroughs like their Streaming Multi-layer Truncated Attention model (SMLTA) in 2019, which improved recognition accuracy by 15% while maintaining computational efficiency2.

Recent data shows Baidu’s ERNIE model now processes over 1.5 billion daily API calls—a 7.5x increase—demonstrating significant adoption growth as their AI capabilities mature3.

2️⃣ Intensifying competition drives rapid AI model releases among China’s tech giants

Baidu’s launch reflects the accelerating competitive dynamics in China’s AI landscape, where rivals are rapidly releasing new models to establish market leadership.

The Chinese large language model market is projected to grow at a compound annual growth rate of 37.6% from 2024 to 2032, driving an aggressive race for market share among Baidu, Alibaba, and Tencent4.

Tencent recently countered with its upgraded reasoning model, Hunyuan T1, and announced plans to significantly increase capital expenditure on AI infrastructure4.