DeepSeek begins limited-time beta of V4.1 Flash multimodal model
DeepSeek has commenced a limited-time beta test of its V4.1 Flash multimodal model. The new architecture natively supports multimodal capabilities and is designed for stronger performance.
DeepSeek has launched a limited-time beta program for its V4.1 Flash multimodal model. This interim model features a new architecture and natively supports multimodal capabilities, allowing it to process and generate content across different data types.
The company stated that the model offers stronger performance and faster generation at a lower cost compared to previous iterations. However, DeepSeek has not presented it as a formal product release at this stage.
Developers can access the beta through DeepSeek's existing API by selecting deepseek-v4.1-flash-expires-on-0910. The beta pricing mirrors that of the V4 Flash model, with each account supporting up to 20 concurrent requests.
The V4.1 Flash model is scheduled to be taken offline on September 10. This beta period allows DeepSeek to gather user feedback and evaluate the model's performance in real-world conditions before a potential wider rollout.