DeepSeek Opens Search, Files and Voice APIs

Resumen

Pixel’s announcement covers DeepSeek opening its app capabilities as APIs, including web and deep search, file and URL reading, safety checks, OCR, speech-to-text and text-to-speech. Available through chat integrations or standalone tools, these pay-per-use options could make it cheaper for developers to add useful AI features to their projects.

DEV Community

DeepSeek Just Opened Its Whole App as an API — Search, Files and Voice for Pennies

If you have used DeepSeek's consumer app, you know the feeling: ask a question, it searches the web, reads the PDF you attached, and answers with sources attached. Until now, that whole experience lived behind the app's front door. On 13 November 2026, DeepSeek opened the door — the tools built into its app are now APIs any developer can call. This is not another model release. There is no new flagship to benchmark and no leaderboard to argue about. It is something quieter and, for people who build things, more useful: the capabilities that made DeepSeek's app feel complete are now rentable by the request, at prices that make experimentation almost free. What actually shipped There are two layers to the announcement, and it is worth keeping them separate. Built-in tools inside chat. When you send a chat request to the deepseek-flash model (DeepSeek-V4.1-Flash-1017), you can now switch on the same tools the consumer app uses: web search, deep (multi-step) search, file input, URL reading, and content safety checks. They work across the Messages, Responses and Chat Completions APIs, so you are not locked into one integration style. DeepSeek runs the tools server-side and assembles the results into the model's context — you send a question, the model decides when to search or read, and the answer comes back with the work already done. Standalone tool APIs. If you would rather use the tools without the chat model, they are available on their own: Search and Fetch, OCR, speech-to-text (ASR) and text-to-speech (TTS). You can wire them into your own pipeline, your own model, or your own product, and pay only for what you use. The pricing — and why it matters All prices below are DeepSeek's own published figures, reported by the vendor, not independently verified by us. The pattern is the story: every line is cheap enough that cost stops being the reason not to try something. Tool Price (vendor-reported) Search and Fetch $0.75 per 1,000 successful requests OCR $0.225 per 1,000 pages Speech-to-text (ASR) $0.116 per hour — currently 50% off at $0.058 Text-to-speech (TTS) $0.45 per 10,000 characters — currently 50% off at $0.225 deepseek-flash tokens $0.30 per 1M input / $1.20 per 1M output at peak; half that off-peak There is no monthly fee. It is pay-as-you-go, and the off-peak discount on the flash model means a lot of real workloads — overnight batch jobs, background research, queued document processing — run at half the already-low headline rate. To put the voice numbers in plain terms: at the current 50%-off rate, an hour of transcribed audio costs under six cents, and ten thousand characters of spoken output costs about twenty-two cents. That is the kind of pricing where a voice feature stops being a premium tier and starts being a default. Voice is the real news Search and file-reading are useful, but they are extensions of things DeepSeek already offered. Voice is genuinely new. Before this announcement, there was no public DeepSeek endpoint for listening or speaking at all — if you wanted a DeepSeek-powered voice product, you had to bolt on somebody else's speech stack and pay somebody else's prices. Now the loop closes inside one provider: transcribe the user's speech, reason over it with a cheap flash model that can search and read files, and speak the answer back. That is the full shape of a voice assistant, and it arrives at a moment when every other major provider charges noticeably more for the same pieces. We have not tested voice quality hands-on, but the price alone will pull a wave of builders in to find out. A demo client, open-sourced DeepSeek also open-sourced a demo chat client on GitHub that rebuilds the consumer app one-to-one, including importing history exported from the DeepSeek app. It is a working reference implementation, not a toy example: you can see exactly how the tools are meant to be wired together, fork it, and have something real running the same afternoon. What this is not Some context, because...
DeepSeek Opens Search, Files and Voice APIs
Powered by Readybot.io

Lo último de Pixel

Incident Response: Look Beyond the Red Alert

Pixel’s DEV Community announcement explores how teams can fixate on one failing service while overlooking the real cause of an outage, such as a full …

Sach: AI Fact-Checks WhatsApp Forwards in India

Pixel’s announcement introduces Sach, a tool that checks forwarded messages in English, Hindi, Telugu, and 14 other Indian languages against official sources. It gives cited …

Java 25 Removes SecurityManager: What to Fix

Pixel’s announcement explains how Java 24’s JEP 486 permanently disabled the SecurityManager, breaking test workarounds that intercept System.exit. It recommends returning exit codes or throwing …