November 20, 2026

Baidu Releases ERNIE 5.5 Flash Base, a 280B MoE Model

Baidu has published ERNIE 5.5 Flash Base, a mixture of experts model with 280 billion total parameters and 14 billion active per token.
Baidu Releases ERNIE 5.5 Flash Base, a 280B MoE Model

Baidu has released ERNIE 5.5 Flash Base, a new large language model built on a mixture of experts architecture, according to a report by Marktechpost. The model totals 280 billion parameters but activates only 14 billion per token during inference, a design intended to balance model scale with computational efficiency.

The mixture of experts approach allows the model to route each input through a subset of specialized sub networks rather than engaging all parameters at once. This structure is increasingly common among large model developers seeking to offer strong performance while keeping inference costs and latency manageable compared to dense models of similar total size.

Marktechpost identifies the release as a base model, suggesting it is intended as a foundation for further fine tuning and downstream task adaptation rather than a fully instruction tuned consumer-facing product. Base model releases typically target developers and researchers who build specialized applications on top of the underlying architecture.

The release adds to Baidu’s ERNIE line of large language models, which the company has developed as part of its broader push in generative AI. Baidu has been expanding its ERNIE family alongside competition from other Chinese AI labs and global developers, as mixture of experts architectures become a common strategy for scaling model capability without proportional increases in inference cost.

Marktechpost’s report did not include specific benchmark scores or licensing terms for the model at the time of publication. Additional technical details, including training data composition and evaluation results, were not specified in the original source material.

Based on reporting by www.marktechpost.com.

Leave a Reply

Your email address will not be published. Required fields are marked *