DeepSeek-V4-Flash-Vision-Exp Lands on B.AI, Fully Free to Access, With Extra Perks Added
TechFlow reports that on August 23, the B.AI platform launched DeepSeek's new multimodal experimental flagship model, DeepSeek-V4-Flash-Vision-Exp. The official API team has already completed full integration and made it freely available to all users. Building on V4-Flash's top-tier pure text and code Agent capabilities, the model officially unlocks native image understanding capabilities, with its comprehensive multimodal Agent performance now approaching that of the Claude Opus-4.8 flagship. The new model supports mixed text-image input and a 1M ultra-long context window, effortlessly handling demanding scenarios such as multimodal PPT generation, complex UI reconstruction, and frontend visual effects. Image inputs are billed per token, costing a maximum of just 384 tokens per image. Users can now log in to the B.AI platform to freely invoke this latest multimodal model capability with zero barriers, accelerating the practical deployment of top-tier computing power for creativity and productivity.





