January 27th, 2025

New!

✨ New Model: DeepSeek R1

On Friday we added DeepSeek V3—a capable and extremely economical AI model. Today we’re adding its super smart sibling: DeepSeek R1. In our tests, DeepSeek R1 has reasoning capabilities comparable to OpenAI’s o1 model, but at a fraction of the cost.

This model works a bit differently than others. Before working on a task, it will spend up to a few minutes “thinking,” so you’ll likely see a significant delay before you it begins writing (sometimes minutes). The providers for this model are also in flux to meet demand, so there may be stability issues in the short term.

Despite these issues, we’ve found the prose quality and instruction following capabilities of this model to be stellar for its price, which is why we wanted to get it into your hands ASAP.

DeepSeek R1 is now available in the Write Button, Chapter Generator, and Plugins. Please give us feedback on how it’s performing! This model was just released a few days ago, so our team is still in the process of discovering and optimizing its capabilities.