Baidu has launched Ernie Bot 5.0, a multimodal large language model with 2.4 trillion parameters.

The model integrates text, images, audio, and video in a single architecture, aiming to improve understanding and generation across complex tasks.

Baidu claims Ernie Bot 5.0 outperformed several international models, including Gemini-2.5-Pro and GPT-5-High, on more than 40 benchmarks.

It uses a Mixture-of-Experts architecture, activating less than 3% of parameters during inference to enhance efficiency.

Baidu collaborated with 835 experts from various fields to guide the model’s development, focusing on logic, professional knowledge, and creative output.

Ernie Bot 5.0 is available to individual users through Baidu’s app and website. Businesses and developers can access it via Baidu AI Cloud’s Qianfan platform.

The company aims to promote AI adoption across industries.

🔗 Source: Pandaily

Recent Baidu developments