O3, developed by OpenAI, has been ranked the best AI model for answering scientific questions across various domains according to the new SciArena platform. This platform ranked 23 large language models based on answers provided to scientific questions submitted by researchers. The evaluation included responses from 102 researchers who voted on answer quality. O3 excelled particularly in natural sciences, healthcare, engineering, and humanities. SciArena represents a novel initiative in AI benchmarking through crowdsourced feedback, aiding in the assessment of LLM performance on science-related tasks.
O3, an AI model from OpenAI, achieved the top rank among 23 LLMs in answering scientific questions across multiple fields, evaluated on the SciArena platform.
The evaluation involved 102 researchers who voted on the quality of answers, leading o3 to be recognized for its detailed citations and nuanced responses.
DeepSeek-R1 and Google's Gemini-2.5-Pro followed in rankings, illustrating varying strengths in natural sciences, engineering, and healthcare among competing models.
SciArena represents a new approach to assessing AI performance, incorporating crowdsourced feedback to rank LLMs on their abilities to handle scientific queries.
Collection
[
|
...
]