(Photo/VCG)
China’s high-tech heavyweight integration announced the launch of its new model architecture, Ultramem, on Thursday. This reduced the inference costs of artificial intelligence-driven models by up to 83%.
According to the company’s “S Doubao LLM team, Ultramem enhances inference speed by 2-6 times compared to traditional MOE (mixture) architectures. This technological advancement is the reasoning efficiency of large-scale language models and the inference efficiency. Provides new routes to improve performance.
This move follows the surprising release of Deepseek’s high-performance, cost-effective open-source AI model R1.
Additionally, Baidu Inc revealed on Thursday that its AI chatbot, Ernie Bot, will be available for free from April 1, thanks to upgraded technology and cost reductions. The AI service provides free access to all users on both desktop and mobile platforms, the company said.
Additionally, Baidu has also launched an advanced search feature that is free to use from April 1st.
A small number of features improve inference ability and tool integration to provide expert-level responses, handle multiple tasks, and achieve multimodal inputs and outputs, Baidu said.
