How much do the DeepSeek API tools cost?

DeepSeek charges the built-in tools on top of normal token prices. Its pricing page lists web search and URL reading at $0.75 per 1,000 calls and file parsing or OCR at $0.225 per 1,000 pages. Text to speech is $0.45 per 10,000 characters and speech to text is $0.116 per hour, both at half price for a limited time.

Which DeepSeek models support built-in web search and file input?

DeepSeek's release applies the built-in features to V4.1-Flash, called deepseek-flash in the API. They work on the Messages, Responses and Chat Completions endpoints, and DeepSeek's servers handle the tool calls and context assembly. The standalone Search, Fetch, OCR, ASR and TTS APIs can also be called directly for custom workflows.

Are there limits when DeepSeek web tools or safety checks are turned on?

DeepSeek's chatbot guide says that when the web tools, URL reading, a high safety_check level or conversation titles are enabled, the service ignores several request parameters, such as temperature, top_p, stop and seed. In this mode max_tokens applies to each inference step in a round, not to the round's total output.

Do DeepSeek's safety checks cover an app's legal obligations?

No. DeepSeek says public-facing apps must enable the safety checks, but its guide also says the checks do not replace a developer's own duties. Services must still handle their own content review, large model registration and AI labeling, and follow the DeepSeek Open Platform Terms of Service.

DeepSeek · 2026-11-13 · major

DeepSeek App API — web search, file input and safety checks built into the API

In short

DeepSeek now offers its app's features through the API. deepseek-flash (V4.1-Flash) can search the web, read links, parse files and run safety checks on the server, and new OCR, ASR and TTS APIs are available too.

DeepSeek · 2026-11-13 · 1 source checked

DeepSeek moves the tools behind its chat app into the API, so developers can rebuild the app with a few request flags.

Quick facts

Maker	DeepSeek
Model	deepseek-flash (V4.1-Flash)
Built-in tools	Web search, URL reading, file input, safety checks
Standalone APIs	Search, Fetch, OCR, ASR, TTS
Endpoints	Messages, Responses, Chat Completions
Released	November 13, 2026

Pricing

Web search	$0.75 / 1,000 searches
URL reading	$0.75 / 1,000 fetches
File parsing and OCR	$0.225 / 1,000 pages
Text to speech · $0.225 limited-time	$0.45 / 10,000 characters
Speech to text · $0.058 limited-time	$0.116 / hour of audio

source ↗

What is it?

DeepSeek's November 13 API release gives developers the same features as the DeepSeek App. The deepseek-flash model (V4.1-Flash) gets built-in web search with citations, a deeper search mode inside thinking, URL reading, file input for PDF, Office and image files, and input and output safety checks.

How does it work?

Developers turn the features on in the request: the deepseek_web_search and deepseek_read_url tools, a safety_check level, and thinking mode. DeepSeek's servers run the tool calls and assemble the context, and this works on the Messages, Responses and Chat Completions endpoints. OCR, speech-to-text and text-to-speech are also offered as separate APIs for custom pipelines.

Why does it matter?

Teams no longer need their own search provider, scraper or document parser to build a DeepSeek-based assistant. Tool calls are billed per use on top of tokens, so the cost of a full chatbot is easier to plan. DeepSeek says public-facing apps must enable the safety checks.

Who is it for?

developers building chatbots and agents on DeepSeek

Frequently asked questions

How much do the DeepSeek API tools cost?: DeepSeek charges the built-in tools on top of normal token prices. Its pricing page lists web search and URL reading at $0.75 per 1,000 calls and file parsing or OCR at $0.225 per 1,000 pages. Text to speech is $0.45 per 10,000 characters and speech to text is $0.116 per hour, both at half price for a limited time.
Which DeepSeek models support built-in web search and file input?: DeepSeek's release applies the built-in features to V4.1-Flash, called deepseek-flash in the API. They work on the Messages, Responses and Chat Completions endpoints, and DeepSeek's servers handle the tool calls and context assembly. The standalone Search, Fetch, OCR, ASR and TTS APIs can also be called directly for custom workflows.
Are there limits when DeepSeek web tools or safety checks are turned on?: DeepSeek's chatbot guide says that when the web tools, URL reading, a high safety_check level or conversation titles are enabled, the service ignores several request parameters, such as temperature, top_p, stop and seed. In this mode max_tokens applies to each inference step in a round, not to the round's total output.
Do DeepSeek's safety checks cover an app's legal obligations?: No. DeepSeek says public-facing apps must enable the safety checks, but its guide also says the checks do not replace a developer's own duties. Services must still handle their own content review, large model registration and AI labeling, and follow the DeepSeek Open Platform Terms of Service.

Try it

POST https://api.deepseek.com/chat/completions with model "deepseek-flash" and tools [{"type": "deepseek_web_search"}]