Faraday 27B model claims to outperform Claude Opus 4.8 and GPT-5.5 in replicating papers

neal_lathia · x · 2026-08-16

Inherent Labs introduced Faraday, a 27B-parameter AI Scientist designed with scientific intuition. Trained via long-horizon reinforcement learning, it claims to outperform Claude Opus 4.8 and GPT-5.5 on the task of replicating research papers.

Related event: 27B Model Faraday Outperforms Claude Opus 4.8 in Paper Reproduction(16 posts)→

Original post →

More from Models

Models channel →