Just a week after Bytedance’s amazing launch of Omnihuman-1, Baidu, an AI model that turns photos into realistic human videos, is preparing to release the next generation of artificial intelligence models later this year .
According to an exclusive report from CNBC, Baidu’s new model, Ernie 5, includes the ability to process and convert between text, video, images and audio. Known as the “foundation model,” Ernie 5 is expected to significantly improve on these different forms of handling, but certain details remain under the wrap.
“China’s Baidu is planning to release the next generation of artificial intelligence models later this year,” CNBC reported, citing sources familiar with the issue.
This release comes with a heated competition in China’s AI sector. Startups like Deepseek are making waves, especially after launching inference models comparable to Openai’s GPT, but at a small cost.
Despite being an early player in AI following the 2022 ChatGPT debut, Baidu has not seen extensive adoption of Ernie’s major language models. The company claims that the current version of the Ernie 4 is comparable to the GPT-4, but continues to follow competitors such as Bytedance’s Doubao Chatbot and the new Deepseek in terms of user count.
Baidu CEO Robin Li worked on this at a meeting in Dubai, pointing to the rapid rise of Deepseek as an example of what unpredictable innovation could turn out. “You don’t know when or where you came from,” he said. Li also highlighted the ongoing need for investment in data centers and cloud infrastructure, despite the new model promises greater cost-effectiveness.
Basic models like Ernie 5 are designed to handle a variety of tasks, from text and image generation to engage in natural conversations. Baidu’s update is as Chinese tech companies compete to compete with Openai and other US-based companies. In January, Deepseek launched the global technology market with the release of its open source AI model, praised its inference capabilities and significant costs compared to ChatGpt.
“We live in an exciting time… reasoning cost [of foundation models] Li is a World Government Summit in Dubai that can cut it by more than 90% in 12 months. He noted that reducing these costs directly increases productivity, a key driver of innovation.
Baidu was the first Chinese high-tech company to launch Arnie in March 2023 and introduce chatbots like ChatGpt.
As we also reported in 2023, Baidu launched a $145 million venture capital AI fund to back up a generator AI startup focused on content generated by artificial intelligence applications. A few months ago, Baidu launched Ernie Bott, a large-scale language model (LLM), which is powered by its own AI, similar to Openai’s ChatGpt.
Source link