[DeepSeek-V4.1-Flash Officially Launches on B.AI, Native Multimodal MoE Model Opens Dual Access via Web Chat and API] According to TechFlow by DeepTrend, on September 10, the B.AI platform announced the official launch of the newly released native multimodal MoE model, DeepSeek-V4.1-Flash. This model features a 552 billion parameter backbone and a pioneering Causal Encoder-Decoder architecture, utilizing asymmetric computational scales for input and output. It significantly reduces operational costs while comprehensively surpassing the previous flagship model. The model supports a 1 million token context and an ultra-large 384K output, specifically designed for long-cycle programming agents, multimodal workflows, and high-concurrency agent systems. Currently, dual access via Web Chat and API is open, allowing developers to experience this cost-effective multimodal model with zero barriers. More exciting benefits are on the way—stay tuned to B.AI's official updates.