DeepSeek R1 may soon become Perplexity's default model, raising daily limit to 500

Perplexity, the popular AI-powered search engine, is moving fast in adopting the DeepSeek R1 model.

The company’s CEO, Aravind Srinivas, is considering making the open-source model Perplexity AI’s default model.

The company has previously added the open-source model to the search engine with a limit of 25 at its launch, and now, it’s adding up to even 500 for paid Perplexity Pro users. Free users can also try it, but only with 5 daily queries for the time being.

Unlike its original Chinese counterpart, DeepSeek R1 on Perplexity has been polished so it’s not heavily censored and is stored in the US instead. It is available on iOS, macOS, web, and Android.

The CEO also says that new reasoning models can make the native assistant on Android a whole lot better, especially for complex tasks. “We will make the native assistant on the phones (Android to begin with) accomplish tasks more reliably. And the Assistant will be free as long as you switch default from Google/Gemini,” he adds.

The arrival of DeepSeek caused a lot of anxiety in the US market, especially for its claims that it built the model using older Nvidia H100 GPUs and just $6 million. But, a lot of companies have started adopting it.

Microsoft has now said that it’s launching DeepSeek R1 to Copilot+ PCs and developers could try it on Azure AI & GitHub. Though, both Microsoft and OpenAI have been investigating whether the Chinese company improperly accessed OpenAI’s technology data through its API, which is a violation of OpenAI’s terms of service.