One month after the initial release of its ERNIE 4.5 and ERNIE X1 models, Chinese tech giant Baidu has launched upgrades to its multimodal foundation model and reasoning model at its annual developer conference in Wuhan, China. The new models, dubbed ERNIE 4.5 Turbo and ERNIE X1 Turbo, boast enhanced multimodal capabilities, strong reasoning, low costs, and are available for users to access on Ernie Bot free of charge.

During his keynote, CEO Robin Li emphasized the importance of empowering developers to build innovative applications without worrying about model capability costs or development tools. Li stated that "without practical applications, neither advanced chips nor sophisticated models hold value." The company plans to integrate both new models into its product ecosystem, including Baidu Search, China's largest search engine, and other offerings.

However, industry analysts have responded with a lukewarm reception to the announcements. Paul Smith-Goodson, vice president and principal analyst for quantum computing, AI, and robotics at Moor Insights & Strategy, pointed out that Baidu's achievement of "illuminating" a cluster of 30,000 self-developed P800 chips is merely a technical advancement, and not exceptional compared to the capabilities of companies like OpenAI, Google, IBM, Anthropic, Microsoft, and Meta.

Smith-Goodson also expressed skepticism about the performance of Baidu's models compared to global leaders, citing the lack of benchmarks and the inferiority of Kunlun chips to US GPUs. He noted that the race to build the first Artificial General Intelligence (AGI) model is heating up, with the US still holding a lead, but China pressing hard to catch up.

Thomas Randall, director of AI market research at Info-Tech Research Group, also expressed a lukewarm response, but acknowledged Baidu's importance in China's competitive AI sector. He noted that Baidu's ERNIE models are one of the few domestically developed Large Language Models (LLMs) that compete with OpenAI/GPT-level models, and the Kunlun chips and new cluster announcement reinforce Baidu's position as a broad provider of hardware and applications.

However, Randall also highlighted the challenges Baidu faces, including immense pressure from emerging startups like DeepSeek, Moonshot AI, and cloud giants like Alibaba. He noted that Baidu remains largely irrelevant in Western countries due to the lack of trust in geopolitics and the decoupling of the US and Chinese tech ecosystems, making Western expansion near impossible.

Despite these limitations, Randall emphasized that Baidu remains strategically relevant globally, pushing the AI race forward. The key takeaway for Western AI companies is that innovation is not US-centric, and that global competition will drive progress in the AI field.

In conclusion, while Baidu's upgrades to ERNIE 4.5 and ERNIE X1 models may not have impressed industry analysts, the company's continued investment in AI research and development is a significant step forward for the Chinese tech giant. As the AI race intensifies, it will be interesting to see how Baidu and other players respond to the challenges and opportunities that lie ahead.