> ## Content Index
> Fetch the complete content index at: https://www.testingcatalog.com/llms.txt
> Use this file to discover other available public pages before exploring further.

# DeepSeek R1 now combines real-time web data with advanced reasoning
- URL: https://www.testingcatalog.com/deepseek-r1-now-combines-real-time-web-data-with-advanced-reasoning/
- Published: 2025-01-22T23:36:51.000Z
- Updated: 2025-01-25T10:47:47.000Z
- Author: Alexey Shabanov
- Tags: DeepSeek, Latest AI News

DeepSeek R1, a reasoning-focused AI model developed by the Chinese AI lab DeepSeek, has introduced a significant new feature: simultaneous web browsing and reasoning capabilities. This enhancement allows the model to access real-time web data while leveraging its advanced reasoning framework, thereby improving its ability to provide accurate and up-to-date responses.

### The New Feature

[DeepSeek R1](https://www.testingcatalog.com/open-source-deepseek-r1-challenges-openais-o1-in-advanced-benchmarks/) now integrates real-time web browsing into its reasoning process. This capability enables the model to fetch and analyze live information from the internet, combining it with its pre-trained knowledge to deliver more comprehensive answers. Users can access this feature via the DeepSeek API or its web-based chat platform by activating the "Deep Thinking" mode. The integration mirrors functionalities seen in models like Perplexity AI, which specialize in combining reasoning with live data retrieval.

> DeepSeek app for iOS now supports semultanious Search with Deep Think, following the web release 👀 [https://t.co/7TiqiinUCm](https://t.co/7TiqiinUCm?ref=testingcatalog.com) [pic.twitter.com/0rbluv6VrR](https://t.co/0rbluv6VrR?ref=testingcatalog.com)
> 
> — TestingCatalog News 🗞 (@testingcatalog) [January 24, 2025](https://twitter.com/testingcatalog/status/1882828734816018720?ref%5Fsrc=twsrc%5Etfw&ref=testingcatalog.com)

### About DeepSeek

[DeepSeek](https://www.testingcatalog.com/tag/deepseek/) is an emerging player in AI, known for its focus on reasoning models. The R1 series, including R1-Zero and R1, employs innovative training methods such as reinforcement learning without supervised fine-tuning (R1-Zero) and a hybrid approach combining reinforcement learning with supervised fine-tuning (R1). These models are open-source under the MIT license, emphasizing transparency and community collaboration. The company has also released smaller distilled versions of R1 to make advanced reasoning capabilities accessible on less powerful hardware.