Baidu launches new AI models, touts superiority to DeepSeek, OpenAI on benchmarks
Baidu has released two new AI models, Ernie 4.5 and Ernie X1, claiming superior performance to DeepSeek and OpenAI's models on certain benchmarks. This move aims to regain ground in China's competitive AI market, where Baidu's early lead has been challenged by other tech companies.
Ernie 4.5, a multimodal foundational model, reportedly surpasses OpenAI's GPT-4o in image, audio, and video capabilities, as well as DeepSeek V3 in text-based tasks, according to Baidu. While benchmark results for Ernie X1, a multimodal reasoning model, were not provided, Baidu claims it offers comparable performance to DeepSeek R1 at half the cost, with API access priced at 2 yuan per million token inputs and 8 yuan per million token outputs.
via SCMP Full Text Feed
Baidu has released two new AI models, Ernie 4.5 and Ernie X1, claiming superior performance to DeepSeek and OpenAI's models on certain benchmarks. This move aims to regain ground in China's competitive AI market, where Baidu's early lead has been challenged by other tech companies.
Ernie 4.5, a multimodal foundational model, reportedly surpasses OpenAI's GPT-4o in image, audio, and video capabilities, as well as DeepSeek V3 in text-based tasks, according to Baidu. While benchmark results for Ernie X1, a multimodal reasoning model, were not provided, Baidu claims it offers comparable performance to DeepSeek R1 at half the cost, with API access priced at 2 yuan per million token inputs and 8 yuan per million token outputs.
via SCMP Full Text Feed
