Baidu open-sources ERNIE 5.5 Flash Base, a 280B-parameter MoE model

AIBaidu says it has open-sourced ERNIE 5.5 Flash Base, a MoE base model with 280B total parameters and 14B active per token. The model was pretrained on 33.5 trillion tokens with a 128K context length, extendable to 1M tokens during fine-tuning and inference. Baidu reports state-of-the-art results among same-size models across several benchmarks, ahead of DeepSeek-V4-Flash-Base, Hy3-preview-Base, and MiMo-V2.5-Base.

AIBaidu open-sourced ERNIE 5.5 Flash Base on November 5, a MoE base model with 280 billion total parameters and 14 billion active per token. The company says it was pretrained on 33.5 trillion tokens, supports up to 128K context in pretraining and up to 1 million tokens in fine-tuning and inference, and outperforms DeepSeek-V4-Flash-Base, Hy3-preview-Base, and MiMo-V2.5-Base on several benchmarks of its size class.