# DeepSeek exposes app tools and voice features via API

DeepSeek V4.1-Flash now supports web search, link reading, file parsing, and safety checks server-side alongside standalone OCR, ASR, and TTS endpoints.

- Extended capabilities incur separate tool usage fees in addition to token charges.
- Web search supports citations and images in standard mode, as well as multi-step search during deep-thinking reasoning.
- File inputs parse PDFs, Office documents, and images directly into the conversation context.
- Developers deploying public-facing applications are required to enable the API's safety checks.
- DeepSeek published an open-source web chat demo on GitHub that supports importing existing DeepSeek conversation history.

## Why it matters

Developers can now run grounded, multimodal DeepSeek pipelines without having to host their own scrapers, document parsers, or tool orchestration loops.

## Sources

- [DeepSeek: DeepSeek exposes app tools and voice features via API](https://x.com/deepseek_ai/status/2121247896918253875)

---

Summarized by dstilled on 2026-11-13. https://dstilled.ai/story/9fa50e54-0a5b-4b1c-aa2c-e296c69ae947
