DeepSeek puts its app's functions in the API: search, files, link reading and speech
What does the new DeepSeek API offer? Since November 13, 2026 developers can use the functions of the DeepSeek app through the API, on the deepseek-flash model (V4.1-Flash). They cover web search with citations, search during reasoning, file input, reading links in a message and a safety check. A search and a page fetch each cost $0.75 per thousand, on top of tokens.
We read the announcement in the API documentation and DeepSeek's pricing page. The topic also appeared in the trending briefing of November 14.
What happened
DeepSeek calls the release the "Full-Suite App API". The tools run on DeepSeek's side: the server makes the tool calls and assembles the context. It works across three API formats: Messages, Responses and Chat Completions. There are five built-in functions:
- Web Search (
deepseek_web_search): searches the web or images, with citations, in non-thinking mode. - Deep Thinking Search: searches and reads pages during multi-step reasoning.
- File Input: turns PDFs, Office files and images into conversation context.
- URL Reading (
deepseek_read_url): fetches links in the user's message. - Safety Checks: screens input and output and returns a refusal indicator when it detects risk.
There are also standalone APIs for Search, Fetch, OCR, speech recognition (ASR) and speech synthesis (TTS). DeepSeek offers a demo, and the code is on GitHub as deepseek-ai/chat-demo.
Pricing
The deepseek-flash model runs on DeepSeek-V4.1-Flash-1017, with a context of 1 million tokens and up to 384,000 output tokens. Token prices depend on the time of day: off-peak rates are half. Peak hours are Monday to Friday, 01:00 to 04:00 and 06:00 to 10:00 UTC, excluding Chinese public holidays.
| Item | Unit | Price |
|---|---|---|
| Input, cache miss | per million tokens | $0.15 off-peak / $0.30 peak |
| Output | per million tokens | $0.60 off-peak / $1.20 peak |
| Search (text, image) | per 1,000 searches | $0.75 |
| Web fetch | per 1,000 fetches | $0.75 |
| File parsing and OCR | per 1,000 pages | $0.225 |
| Text to speech | per 10,000 characters | $0.45; temporarily $0.225 |
| Speech to text | per hour of audio | $0.116; temporarily $0.058 |
The pricing page gives no end date for the temporary 50 percent discount on speech. Input that is already cached costs far less; we explained how that works in our piece on context caching. The old model names deepseek-v4-flash and deepseek-v4-flash-vision-exp are still accepted, but those models are retired: such requests go to V4.1-Flash-1017 at the Flash price.
What this means
If you build an agent or chatbot, you no longer need a separate search service, scraper or OCR package: it all sits in one call, priced per use. That makes it easier to build an agent that looks things up by itself, as described in our overview of agentic AI. Files and speech in the same API fit the move towards multimodal models. If your configuration still uses an old model name, you get the newer model without changing anything; test whether that changes how your app behaves.
DeepSeek requires the safety check for public-facing apps. The developer remains responsible for moderation, for model registration where it applies and for labelling AI-generated content. In the EU, the transparency rules of the AI Act come on top, which we explained in our piece on the AI Act.
What we did not check
We did not call the API ourselves or try the demo. For Deep Thinking Search, File Input and Safety Checks the announcement gives no parameter name. We did not verify comparisons with other providers' search prices.
Checked on November 14, 2026.
This article was written with AI (Claude) and checked against the sources below.
Sources
- DeepSeek, Full-Suite App API Release (November 13, 2026): api-docs.deepseek.com
- DeepSeek, Models & Pricing: api-docs.deepseek.com
- DeepSeek, chat-demo on GitHub: github.com
- @deepseek_ai on X (signal): x.com
