On November 13, DeepSeek said the capabilities behind the DeepSeek App are now available through its API. V4.1-Flash (model name deepseek-flash) supports server-side web search (including image search with citations), deep thinking search, file input for PDF, Office and image files, automatic reading of links in user messages, and input/output safety checks across the Messages, Responses and Chat Completions protocols. OCR, speech-to-text (ASR) and text-to-speech (TTS) also launch as standalone APIs. Tool calls are billed on top of token charges: Search and Fetch cost $0.75 per 1K requests, OCR $0.225 per 1K pages, ASR $0.058 per hour and TTS $0.225 per 10K characters (both at a limited-time 50% discount). DeepSeek also open-sourced an MIT-licensed chat front end that can import DeepSeek chat history. Public-facing apps must enable safety checks.
Key Takeaways
- ✓All three protocols are supported: Messages (/anthropic/v1/messages), Responses (/responses) and Chat Completions (/chat/completions), all with the deepseek-flash model
- ✓Turn features on with the deepseek_web_search and deepseek_read_url tools and the safety_check and title parameters; tool calls and context assembly run server-side
- ✓Tool prices (vendor figures, on top of tokens): Search and Fetch $0.75 per 1K requests, OCR $0.225 per 1K pages, ASR $0.058 per hour, TTS $0.225 per 10K characters (ASR and TTS at a limited-time 50% off)
- ✓The open-source demo deepseek-ai/chat-demo is TypeScript under MIT, imports DeepSeek chat history and keeps your API key only in the browser
- ✓The launch post drew about 117K views and 2,100+ likes in two hours; DeepSeek has published no search-quality or OCR/ASR accuracy benchmarks
Key Decision Metrics at a Glance
Turn your technical choice into a development budget
Compare 40 dev plans & simulate token costs vs $20/mo subscriptions
Project Links & Resources
Direct AccessIn-Depth Technical Analysis
Background and pain points
Building a chat product that feels as capable as an official app was never mainly about the model. The hard part is everything around it: web search, page reading, PDF parsing, content safety, and voice in and out, each needing a separate vendor and hand-built context. DeepSeek already opened standalone Search and Fetch APIs on October 17; now it puts the app's whole feature set straight into the chat APIs.
Architecture highlights
V4.1-Flash (deepseek-flash) can turn on built-in tools across the Messages, Responses and Chat Completions protocols. With thinking off, deepseek_web_search runs web or image searches and cites sources; with thinking on it becomes deep thinking search, searching and reading links across multiple reasoning steps. deepseek_read_url fetches links in user messages, files get plain-text extraction or layout-aware parsing, and safety_check screens inputs and outputs and returns a refusal marker. Tool calls and context assembly run server-side; the client only passes returned blocks back unchanged.
Benchmarks and data
DeepSeek has not published search-quality or OCR/ASR accuracy benchmarks for this launch. One reference point: in late October, Artificial Analysis found the DeepSeek Search API at the same $0.75 per 1K price lifted its intelligence score by 36 points. Prices are vendor figures: OCR $0.225 per 1K pages, ASR $0.058 per hour and TTS $0.225 per 10K characters at a limited-time discount.
Getting started
The fastest way in is the official demo at cdn.deepseek.com/chat-demo: paste an API key and try it, with source at GitHub deepseek-ai/chat-demo. For your own app, follow the Build a Chatbot guide, add the two built-in tools and enable safety_check. Public-facing apps must enable safety checks, and developers still own content moderation, model registration and AI-content labeling duties.
Benchmark side-by-side against alternatives, or calculate monthly token cost vs subscription break-even.
Discussion & Comments
0Sign in to join the discussion
Connect with AI developers to exchange benchmark insights.