AI BriefingDeepSeek-R1-Lite-Preview Released: More Powerful Reasoning Ability
Key point
DeepSeek-R1-Lite-Preview showed o1-preview-level performance on AIME and MATH.
1 / 2
Details
DeepSeek-R1-Lite-Preview has been released. It claims o1-preview-level performance on AIME and MATH, with showing the reasoning process in real time being the key feature.
It can be tried right now at chat.deepseek.com. An open-source model and API are also planned to be provided soon.
Inference scaling laws, where the AIME score steadily rises as reasoning length increases, were also presented.
This summary was generated automatically by AI. Check the original for the author's claims and context. Copyright belongs to the original author.
Our guide explains how the AI works. Report summary errors, attribution issues, or removal requests via Contact.
Related
Jevper Released: A Jev Classification Wrapper for OpenAI-Compatible Models with Automatic Logprobs Detection
Rootless Linux Sandbox 'Drop' Released for Isolating AI Agents and Third-Party Programs