Are you interested in staying up to date with the latest industry-leading AI coverage? Join our daily and weekly newsletters for exclusive content and updates. Learn more by subscribing now!
DeepSeek, a cutting-edge AI project from Chinese quantitative hedge fund High-Flyer Capital Management, is dedicated to releasing high-performance open-source technology. The project has just unveiled its latest large language model, the R1-Lite-Preview, which focuses on reasoning and is currently only accessible through the DeepSeek Chat web-based AI chatbot.
Renowned for its contributions to the open-source AI community, DeepSeek’s new release aims to provide advanced reasoning capabilities to the public while upholding its commitment to transparent and accessible AI.
The R1-Lite-Preview is already making waves with its performance, which rivals and sometimes exceeds that of OpenAI’s o1-preview model. Similar to the model released in September 2024, DeepSeek-R1-Lite-Preview showcases “chain-of-thought” reasoning, revealing the various thought processes it undergoes to respond to user queries and inputs.
Despite some of the thought processes seeming unconventional or even incorrect to humans, DeepSeek-R1-Lite-Preview overall delivers remarkably accurate responses. It can even tackle “trick” questions that have stumped other powerful AI models like GPT-4o and Claude’s Anthropic family. Examples include questions like “how many letter Rs are in the word Strawberry?” and “which is larger, 9.11 or 9.9?” Check out the screenshots below from tests conducted on DeepSeek Chat:
A New Approach to AI Reasoning
Designed to excel in tasks requiring logical inference, mathematical reasoning, and real-time problem-solving, DeepSeek-R1-Lite-Preview surpasses OpenAI o1-preview-level performance on benchmarks like AIME and MATH.
