[models]Nov 6, 2026By ByteBulletin EditorBaidu releases ERNIE-5.5-Flash-Base weightsThe 280B-parameter sparse MoE model offers a 1M-token context and novel attention mechanisms designed for efficient fine-tuning.
