DeepSeek said its app tools are now available through its API
DeepSeek published a product update dated November 13, 2026, saying the capabilities behind the DeepSeek App are now available through the DeepSeek API. The official documentation pages give that calendar date. Advance Intelligence recorded the company's public post at 9:24 AM ET the same day. WebProNews published a detailed account on November 14, 2026, at 9:56 AM ET. This is a company product announcement, not a peer-reviewed paper and not an independently audited benchmark or pricing audit.
Real, reported: based on the official documentation and the contemporaneous coverage cited below. Capability lists, tool names, and prices are company claims unless another source describes a test of its own.
What DeepSeek announced
The official post is titled "The full DeepSeek App experience, now via API." It says: "Search, files, links, safety, and voice. The capabilities behind the DeepSeek App are now available through the DeepSeek API. Enable built-in features with a few parameters, or call standalone APIs such as OCR, ASR, and TTS. Build richer conversations or recreate the full DeepSeek App in your own application." The change log, also dated November 13, 2026, puts the same point in the company's own words: "Today we open these capabilities to developers." It also says that in the DeepSeek App, including the web version, "web search, file input, URL reading, voice interaction and safety work together with model inference to form the conversation experience users know."
Advance Intelligence recorded the official account's post at 9:24 AM ET on November 13 and, at the time it captured the thread, listed about 1 million views, 437 replies, 1,114 reposts, and 11,255 likes on that post. Those engagement figures are Advance Intelligence's recorded counts, not an independent audit of the platform.
According to the official post, V4.1-Flash (deepseek-flash) now supports those built-in features across the Messages, Responses, and Chat Completions APIs, "with tool calls and context assembly handled server-side." The November 13 change log is more specific about the switches. With thinking disabled and the deepseek_web_search tool enabled, "the model runs web or image searches as the question requires and answers with citations." With thinking enabled and the same tool on, "the model searches the web and reads links on its own over multiple reasoning steps." File input covers PDFs, Office files, images, and more; "after plain-text extraction or rich-layout parsing, they are added to the conversation context for the model to answer from." With deepseek_read_url enabled, "the model detects links in the user message, fetches the pages and answers with them in context." On safety, the change log describes "basic risk control": "the model input and output are reviewed, and a refusal marker is returned when a risk is found." Developers enable search with deepseek_web_search and URL reading with deepseek_read_url. "Public-facing apps must enable safety checks," the official post says. "Developers remain responsible for content moderation, applicable model registration, and AI-generated content labeling."
The same post, and the change log, say Search, Fetch, OCR, ASR, and TTS are also available as standalone APIs. DeepSeek also pointed to an open-source frontend demo that can import DeepSeek conversation history, and to a "Build a Chatbot" guide. Advance Intelligence, describing the official documentation, wrote that "the service runs the searches and URL reading, while the client shows the returned blocks and passes them back unchanged," and that the open-source demo "stores the API key, conversations and uploaded images only in the browser." WebProNews quoted the official "tool calls and context assembly handled server-side" language and reported that, with thinking mode on, the search tool can make "up to about 20 tool calls before it answers," attributing that figure to the company's web tools guide. The same outlet, citing the speech documentation, wrote that the speech-to-text endpoint "detects 21 languages but takes clips of no more than 120 seconds and doesn't stream," and that text-to-speech "offers two voices in 10 languages and can stream over WebSocket." Those call-count and speech-endpoint figures are not in the two official announcement pages cited as the primary source, so they remain WebProNews's account of the documentation.
How to read the prices
The official Models & Pricing page lists tool fees on top of standard token charges. Search (text or image) is $0.75 per 1,000 searches, and web fetch is $0.75 per 1,000 fetches, whether a developer calls those APIs directly or the model uses them as tools. File parsing and the OCR API are listed at $0.225 per 1,000 pages. Text-to-speech is listed at $0.45 per 10,000 characters, with a limited-time 50% price of $0.225. Speech-to-text is listed at $0.116 per hour of audio, with a limited-time 50% price of $0.058. The page does not say when those speech discounts end, and it says "Product prices may vary and DeepSeek reserves the right to adjust them."
For the deepseek-flash model itself, the same official page lists peak prices of $0.30 per million input tokens (cache miss) and $1.20 per million output tokens, with off-peak rates at half those amounts. Peak hours are given as 01:00-04:00 and 06:00-10:00 UTC, Monday through Friday, excluding Chinese public holidays. In Eastern Standard Time, the evening window is 8:00-11:00 p.m. Sunday through Thursday, and the morning window is 1:00-5:00 a.m. Monday through Friday, so U.S. daytime hours fall in the off-peak window. Those are list prices published by the company, not a third-party measurement of what a typical app will pay.
WebProNews compared the $0.75 search fee with $10 per 1,000 search calls that it said OpenAI and Anthropic list. It also wrote that DeepSeek "counts each query in a multi-query search call separately," and that Anthropic's web fetch tool, in that outlet's reading, has no per-use fee beyond tokens. That comparison is WebProNews's reading of those companies' pricing pages, which are not cited here as primary sources, and it is not an official DeepSeek claim about rivals.
Our advice (general guidance)
If you build on this API, treat the announcement as a product change to check in the official documentation and pricing page, not as a guarantee of search quality, file-parsing accuracy, or safety-check coverage. Company prices and feature lists can change, and a built-in safety switch is not a substitute for your own review of what an app will send, store, or show to users. Public-facing apps, the company says, must enable those checks and still handle moderation and any labeling rules that apply.
Do not paste live API keys, customer files, or production conversation history into unofficial demo clones or into a message that asks you to "enable" a new DeepSeek tool. Even in the official open-source demo, Advance Intelligence's account of the documentation is that the key and uploaded images sit in the browser, which is not a reason to share a production key in a chat or a copied demo. Use the official documentation and the official pricing page you already trust. These prices and tool descriptions are company claims.
These are editorial precautions, not individualized legal, financial or security advice.
Sources
- The full DeepSeek App experience, now via API (DeepSeek API Docs, dated Nov 13, 2026)
- Change Log, Releasing the full suite of application APIs (DeepSeek API Docs, dated Nov 13, 2026)
- Models & Pricing (DeepSeek API Docs)
- DeepSeek Opens Its App's Search, Files and Voice to Developers (WebProNews, John Overbee, Nov 14, 2026, 9:56 AM ET)
- DeepSeek Brings the Full App Experience to Its API (Advance Intelligence, Nov 13, 2026, 9:24 AM ET)
Published November 15, 2026 by FraudMeNot Team. Label: Real, reported.