Today DeepSeek opened an integrated app API. The company said its DeepSeek app
(including web) pairs online search, file input, link reading, voice interaction
and safety features with model inference to power its conversational experience.
Developers can now enable corresponding parameters in the existing API to obtain
deep-search, file-input and other conversational functions; voice and underlying
capabilities (OCR, ASR, TTS) are provided as standalone endpoints and can be
called singly or in combination, allowing developers to implement complex
conversational features or reproduce the full DeepSeek app within their own
applications.