Chinese technology giant Baidu has released two new artificial intelligence models as it races to catch up with industry-shaker DeepSeek.
Baidu was China’s earliest tech giants to launch a generative AI chatbot amidst the frenzy generated by OpenAI’s ChatGPT, but fell behind as it struggled to get widespread adoption for Ernie, its large language model.
Its latest launches — unveiled on the weekend — include a new reasoning-focused model that it says can rival DeepSeek’s AI model.
Also on AF: Server Fraud Case in Singapore May be Linked to AI Chips, China
“ERNIE X1 delivers performance on par with DeepSeek R1 at only half the price,” Baidu said of one of the new models.
The X1 has “stronger understanding, planning, reflection, and evolution capabilities,” Baidu said, adding that it is the first deep thinking model that uses tools autonomously.
The second model, is its latest foundation model ERNIE 4.5, which Baidu says has “excellent multimodal understanding ability.”
Multimodal AI systems are capable of processing and integrating various types of data including text, video, images and audio, and can convert content across these formats.
ERNIE 4.5 “has more advanced language ability, and its understanding, generation, logic, and memory abilities are comprehensively improved,” Baidu said.
The model’s text capabilities surpass those of DeepSeek V3 and roughly match OpenAI’s GPT-4.5, it added.
ERNIE 4.5 also has “high EQ”, which makes it easy for it to understand network memes and satirical cartoons, Baidu said.
While access for both Ernie and DeepSeek is free for the public, both charge businesses, specifically for access to their programming interface, or APIs. DeepSeek charges $0.55 per million token inputs and $2.19 per million token outputs for its reasoning model.
Baidu’s latest launches are priced at half that — $0.28 per million token inputs and $1.12 per million token outputs, according to the South China Morning Post.
Hong Kong-listed shares of Baidu pared back early gains to close with a marginal gain of 0.44%.
Ups and downs
The new launches come at a tense time for Baidu considering DeepSeek’s launch has roiled not just America’s Silicon Valley, but also intensified already fierce competition in China’s AI industry.
Baidu has struggled to gain widespread adoption for its Ernie LLM, despite claiming performance comparable to OpenAI’s GPT-4. Its biggest challenge came from regulatory scrutiny by Beijing, which until last year said it required content from generative AI to be in line with China’s core socialist values.
Strict scrutiny meant Baidu scrapped one of its earliest public launches of Ernie, disappointing investors who drew comparisons with the free-to-use launch of ChatGPT just months before.
Ernie Bot was finally launched to the public in August 2023 and has since racked up 13 million active monthly users. It ranks behind DeepSeek, which has 22.2 million daily active users, and TikTok-parent ByteDance’s Doubao which has 17 million.
In a bid to catch up with DeepSeek, Baidu said last month it would make Ernie Bot free to use for all users starting April 1. It also announced plans to make the Ernie 4.5 series open-source from June 30, like DeepSeek.
DeepSeek’s open-source model was among the primary drivers of its popularity, and Baidu CEO Robin Li acknowledged that last month.
“If you open things up, a lot of people will be curious enough to try it. This will help spread the technology much faster,” he said at an event in Dubai.
- Reuters, with additional editing and inputs from Vishakha Saxena
Also read:
China’s Xi Meets Top Tech Chiefs, Baidu Shares Slip as CEO Absent
China Tightens Focus on Tech Innovation, ‘New Productive Forces’
China Needs Tech Self-Reliance to Avoid Being Strangled: Xi
Baidu Claims Its Latest Ernie AI Is a Match for ChatGPT
‘Let’s Change the Subject’: Baidu’s Ernie Ducks Queries on Xi
Baidu Sues Apple, App Developers Over Ernie Bot Imposters
More Concern on DeepSeek’s Data Policies After ByteDance Link
Investors Pour Money Into China Tech Stocks Amid DeepSeek Fever
Curbing Chinese Access to US AI Models May be Hard: Analysts
Meta Set up ‘War Rooms’ to Study DeepSeek AI Model – Fortune
DeepSeek Breakthrough or Theft? US Probes ‘AI Data Breach’
