DeepSeek has opened the tools behind its consumer chat app to outside developers through its API. Announced on November 13, the release lets the deepseek-flash model search the web, read links and uploaded files, run safety checks and handle voice — the same feature set app users get, now switchable inside an API call. Why it matters: building a search-enabled, file-reading assistant used to mean stitching together several vendors; DeepSeek is now selling that whole stack from one key.
November 14, 2026
What happened
DeepSeek announced the change in its API change log dated November 13 and in a thread on X, describing it as the full DeepSeek App experience, now via API. According to DeepSeek’s documentation, developers turn the new features on with parameters in existing requests across three interfaces: the Messages, Responses and Chat Completions APIs. The built-in tools work with the deepseek-flash model, which currently points to DeepSeek-V4.1-Flash-1017. DeepSeek’s servers run the tool calls themselves, so developers do not have to wire up their own search and parsing loop for the built-in path.
What developers can switch on
With the deepseek_web_search tool enabled and thinking disabled, the model searches the web or images and answers with citations; with thinking enabled, the same tool can search and read pages across several reasoning steps. A separate deepseek_read_url tool fetches links that appear in a user’s message. File input accepts PDF, Word, PowerPoint and image files, which are converted to text or parsed for layout before being added to the conversation. Other options include a safety check that reviews inputs and outputs and returns a refusal marker when it finds a risk, automatic conversation titles, and user-context metadata such as the current date, location and language.
Several capabilities are also sold as standalone APIs — Search, Fetch, OCR, speech recognition (ASR) and text-to-speech (TTS) — for teams that want one piece without the chat model. DeepSeek also published an open-source demo chat client on GitHub that uses the new features and can import conversation history exported from the DeepSeek app; the company describes it as for demonstration only.
What it costs
The tools are billed separately from tokens. DeepSeek’s documentation, as reported on November 13, lists Search and Fetch at $0.75 per 1,000 successful requests, OCR at $0.225 per 1,000 pages, ASR at $0.058 per hour of audio at a limited-time 50% discount (list $0.116), and TTS at $0.225 per 10,000 characters, also at a limited-time half price (list $0.45). Failed searches and empty results are not billed. The deepseek-flash model itself costs $0.30 per million uncached input tokens and $1.20 per million output tokens at peak hours, with off-peak rates at half that. DeepSeek gives no end date for the voice discounts.
Why it matters
This moves DeepSeek from model maker toward tool platform — closer to the hosted-tool offerings of larger U.S. providers, where search, document parsing and voice ship alongside the model. For small teams, the appeal is speed: one provider, one key, and the same product DeepSeek runs for its own users. The catches are the ones DeepSeek itself flags. Public-facing apps are expected to enable the safety check, and developers still handle their own content review, large-model registration and AI labeling duties. Open questions include how long the voice discounts last and how the server-side tools perform at scale — neither source we read had tested the APIs.
Frequently asked questions
Which DeepSeek model supports the new API tools?
The built-in tools work with deepseek-flash, which currently points to DeepSeek-V4.1-Flash-1017, across the Messages, Responses and Chat Completions APIs.
Do the new tools cost extra?
Yes. Search, Fetch, OCR, ASR and TTS carry their own fees on top of token charges; for example, Search and Fetch are listed at $0.75 per 1,000 successful requests.
Is the demo chat client free?
The source code is public on GitHub, but it calls DeepSeek’s paid API using the user’s own key, and DeepSeek describes it as for demonstration only.
Sources: DeepSeek API documentation and change log (via AI Newsway), Startup Fortune, Crypto Briefing