DeepSeek API: Build With Its App Tools Now

3D render illustration of a toolbox with search and voice icons, with the words "DEEPSEEK API TOOLS."

DeepSeek's chat app can search the web, read your files, understand your voice and answer from live information. Until now, those abilities lived only inside the app. On 13 November 2026, DeepSeek announced that the whole toolkit is now available through its API.

The release turns the DeepSeek app into building blocks — search, file reading, voice and safety checks — each usable on its own or together. Here is what you get, what it costs, and how to start.

What changed

The update arrived in DeepSeek's API change log on 13 November, announced the same day on X.

At the centre is a new model, deepseek-flash — the DeepSeek-V4.1-Flash-1017 checkpoint — powering the new capabilities across three existing endpoints: Messages, Responses and Chat Completions. Developers switch the features on with parameters in a normal request.

If you already call the DeepSeek API, the new tools arrive as options on requests you already make. Several can also be called as standalone services, without the chat model.

The built-in tools

Five capabilities now work inside chat requests. The deepseek_web_search tool connects the model to live web search results, including image search, and the model answers with citations. A separate deepseek_read_url tool fetches pages linked in a message.

File input accepts PDFs, Word and PowerPoint files, and images. Documents are converted to text or parsed for layout: text_recognition extracts written content, while layout parsing keeps the structure of tables, forms and slides.

Content safety checks screen material for public-facing apps, and automatic conversation titles plus user-context metadata handle the small jobs that make a chat product feel finished. Public-facing apps must include its safety checks.

The standalone APIs

Alongside the built-in tools, five capabilities are also available as standalone APIs, each with its own endpoint and price.

The Search API runs searches through DeepSeek's internal search engine and returns structured results, at up to 100 requests per second. Image search is not currently supported. The Fetch API retrieves a web page's contents and returns it in Markdown.

The OCR API turns images of printed text into machine-readable words — a scanned receipt, a photo of a whiteboard, a PDF of a contract. It accepts files up to 100 MB and documents up to 1,000 pages, returning results page by page. Files processed here are not stored or passed to the chat model.

The ASR API converts speech into text in 21 languages with automatic detection, accepting WAV and Ogg Opus files up to 120 seconds and 8 MB. The TTS API turns written output into spoken audio in 10 languages, with two voices — Stella and Tide — up to 10,000 characters per request, with streaming audio via WebSocket.

What it costs

The tools are billed separately from tokens. DeepSeek lists these prices as of 13 November: Search and Fetch $0.75 per 1,000 requests; OCR $0.225 per 1,000 pages; ASR $0.116 per hour, half price at $0.058; TTS $0.45 per 10,000 characters, half price at $0.225. No end date is given for the voice discounts.

The 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. For a small project the tool fees are the smaller part of the bill; for a search-heavy product, they are the part to watch.

File limits to know

Files up to 100 MB are supported, documents up to 1,000 pages, and a single request can reference up to 20 documents, provided total extracted text does not exceed 8 MB. Formats include PDF, Word and PowerPoint files, and images.

Those limits make the API practical for real document work — contract review, a research assistant, a support bot that answers from your manuals. The 8 MB extracted-text cap is the number to plan around.

How to start

DeepSeek has also released an open-source demo chat client with voice input, spoken replies and history import — the fastest way to see the tools working together.

Before you ship, turn on the safety checks if your app is public-facing — DeepSeek requires it — and test with real volumes, since tool fees sit on top of token prices.

Key takeaways

  • DeepSeek opened the tools behind its chat app to developers through its API on 13 November 2026, powered by the deepseek-flash model (V4.1-Flash-1017).
  • Built-in tools cover web search, URL reading, file input, safety checks, automatic titles and user-context metadata, across the Messages, Responses and Chat Completions endpoints.
  • Five standalone APIs cover Search, Fetch, OCR, speech recognition (ASR) and text-to-speech (TTS), each with its own endpoint and price.
  • Tool prices: Search and Fetch $0.75 per 1,000 requests; OCR $0.225 per 1,000 pages; ASR $0.116 per hour and TTS $0.45 per 10,000 characters, both currently half price.
  • File limits: up to 100 MB and 1,000 pages per document, 20 documents per request, 8 MB total extracted text. Public-facing apps must include safety checks.

Conclusion

DeepSeek's move is simple: the app is now an API. Developers no longer have to choose between DeepSeek's low model prices and the everyday features that make a chat product useful — search, files, voice and safety now come from the same place. The tools are cheap enough to try this week.

FAQ

What is deepseek-flash?

Deepseek-flash is the model powering DeepSeek's new API tools, identified as the DeepSeek-V4.1-Flash-1017 checkpoint. It works across the Messages, Responses and Chat Completions endpoints, at $0.30 per million uncached input tokens and $1.20 per million output tokens at peak hours.

How much do DeepSeek's new API tools cost?

Tool fees sit on top of token prices. Search and Fetch cost $0.75 per 1,000 requests, OCR $0.225 per 1,000 pages, ASR $0.116 per hour and TTS $0.45 per 10,000 characters — both voice services are currently half price, with no end date given.

Do I have to use the safety checks?

If your app is public-facing, yes. DeepSeek requires public-facing apps built on the API to include its safety checks.

Researched with AI assistance and reviewed for accuracy.

Sources

Follow on WhatsApp

Follow Daily Tech & AI on WhatsApp for one tech and AI update every day: Daily Tech & AI WhatsApp channel.

Comments