Maybe you’ve heard the recent AI news about this new open-source model that is outperforming ChatGPT Pro (that’s the $200/month plan).
Maybe you want to take this model for a test drive, but you’ve also heard about privacy concerns with using it through the DeepSeek site, and you’re hesitant.
The good news is that you may already have access to it through Perplexity AI - Pro subscription. Even if you don’t have a Pro subscription, it is $20 per month, so you can subscribe for a month, try out the model, and cancel.
Follow these steps to use DeepSeek R1 with Perplexity Pro -
- Login to Perplexity Pro
- (I tried to use it before logging in, and it wouldn’t let me, so I don’t think that it works with a free account - it could be different by the time you read this, though)
- Click on the Pro slider
- Select Reasoning with R1
- You will know that it took when the slider says “Pro - R1”
Now, you can ask a question like you normally would.
If you haven’t used a reasoning model before (like GPT-o1, Gemini 2.0 deep research, DeepSeek R1, etc.), then the output might look a little weird at first. In this case, DeepSeek R1 will use a lot more sources than usual, identify the research steps that it will take, and identify the reasoning steps that it will take beforehand. It will also take longer than usual, but the result is a more detailed answer.
This is a clip of the R1 search process:
There’s more after this (a lot more) but just wait for it to finish and see if it is valuable for you.
Comments