Loading...
Market
Story
DeepSeek's New AI Model Matches OpenAI's Performance in Key Benchmarks
Nov 20, 2024, 12:26 PM
DeepSeek has released a new AI model named DeepSeek-R1-Lite-Preview, which has demonstrated performance comparable to OpenAI's o1-preview in AI research tests, particularly excelling in the AIME and MATH benchmarks. This model, developed by DeepSeek, a Chinese AI company, showcases reasoning capabilities and provides transparent thought processes in real-time. The model is currently available for testing online, with plans for open-source models and APIs to be released soon. This development signifies a notable advancement in AI research, positioning DeepSeek as a strong competitor to leading AI labs like OpenAI.
View original story
Similar markets
Yes • 50%
No • 50%
Yes • 50%
No • 50%
Yes • 50%
No • 50%
Top 10% • 25%
Top 1% • 25%
Top 5% • 25%
Below Top 10% • 25%
o3 remains the top performer • 25%
Another model surpasses o3 • 25%
o3 ties with another model • 25%
No new models tested • 25%
Yes • 50%
No • 50%
Reinforcement learning • 25%
Search-based reasoning • 25%
Thinking before answering • 25%
Other • 25%
