DeepSeek has made a wider set of its app capabilities available to developers through its hosted API, including web search, file input, URL reading and voice-related services. Its November 13 change log describes these as features around the existing deepseek-flash service, not a new model release.
Developers can enable search and URL reading in supported conversation APIs. DeepSeek says the service handles tool calls and assembles the returned context. It also lists separate search and fetch, optical character recognition, speech recognition and text-to-speech APIs. Search on its own was already opened in October; the November announcement expands the surrounding application suite.
What changes for builders
Hosted retrieval and file parsing reduce the plumbing an application must implement itself, but they also introduce external dependencies. Developers need to account for which files, links and audio are sent to the provider, review data-handling terms and verify citations rather than assume a retrieved page supports a response.
DeepSeek also published a demo client repository for these APIs. Its client code and the hosted model or services have separate terms; an open demo does not make the cloud service local or free. Consult the provider's current pricing and availability before building around it.
Our view: the most useful improvement here is access to the building blocks individually, especially speech and OCR. Whether it is the right stack depends on privacy, regional availability, cost and how accurately it handles the user's actual material. We have not called the APIs or independently tested their performance.
Original reporting; primary sources reviewed November 14, 2026. No independent hands-on test claimed.
