TPSTokens Per Second

ERNIE 4.5 21B A3B

Baidu AI🇨🇳 China
active

Baidu's text-based MoE model with 21B total parameters and 3B activated per token. Exceptional multimodal understanding through heterogeneous MoE structures and modality-isolated routing.

Compare with other models →
Context window120K tokens
Input / 1M tokens$0.07
Output / 1M tokens$0.28

Version History

ernie-4.5-21b-a3b-2025-08-12major

ERNIE 4.5 21B A3B launches as Baidu's efficient MoE text model with 21B params (3B active). 120K context at $0.07/$0.28 per 1M tokens.