A Chinese AI startup, DeepSeek, has introduced a new artificial intelligence model that it claims rivals OpenAI’s o1 in performance while being significantly more affordable and fully open-source.
In a recent research paper, DeepSeek unveiled its flagship AI model, R1, highlighting its advanced reasoning capabilities. The company announced on social media, stating, “DeepSeek-R1 is here. Comparable to OpenAI-o1. Fully open-source with an MIT license—free to modify and commercialize.”
Industry experts believe that this development represents a major breakthrough in terms of both scale and efficiency. Unlike other high-cost AI models like ChatGPT, DeepSeek’s R1 offers similar performance while operating on a much lower budget, potentially reshaping expectations around the computing power required for AI advancements.
Cost-Effective AI with Comparable Performance
According to Shubham Saboo, a senior AI product manager, DeepSeek R1 offers a substantial cost advantage. “DeepSeek R1 is fully open-source and delivers 96.4% cost savings compared to OpenAI o1. While OpenAI charges $60 per 1 million output tokens, DeepSeek R1 does the same for just $2.19,” he shared in a tweet.
Tech entrepreneur Arnaud Bertrand likened the impact of DeepSeek’s model to releasing a smartphone comparable to the iPhone but selling it at a fraction of the cost.
DeepSeek asserts that R1 performs exceptionally well in areas such as mathematics, coding, and reasoning tasks. The company attributes this success to the use of pure reinforcement learning—a technique similar to what powered Google DeepMind’s AlphaZero to master games like Chess and Go.
Open-Source AI for All
Founded in 2023, DeepSeek’s mission is to explore AI’s reasoning potential without relying on supervised data. Their research revealed that an earlier version, R1-Zero, showcased the model’s ability to refine its thinking process and reassess its problem-solving approach over time.
Jim Fan, a senior research manager at Nvidia, expressed excitement over the development, stating that DeepSeek is fulfilling the original mission of OpenAI by making frontier AI research truly open and accessible to all.
One of the most significant implications of DeepSeek’s open-source nature is its potential to make AI-powered applications more affordable and accessible. Bertrand explained that AI services that were previously cost-prohibitive could now become viable.
For instance, summarizing a book using OpenAI’s pricing structure would cost nearly $2 per book. In contrast, DeepSeek’s API would allow the same task to be completed for just $0.07—making it possible to summarize around 30 books for the same cost as a single book with OpenAI.
DeepSeek’s disruptive pricing and open-access approach could reshape the AI landscape, offering developers and businesses more opportunities to build cost-effective AI-driven solutions.
While OpenAI kick-started the AI revolution with the release of ChatGPT in 2022, competition from global players like Google, Meta, and now DeepSeek is rapidly intensifying, leveling the playing field and pushing the boundaries of AI innovation.
