Inherent Claims 'Faraday' Agent Outperforms AI Giants

Inherent, a London-based AI lab founded by former Google DeepMind alumni, claims its new AI agent, Faraday, has outperformed larger frontier models at a specific scientific task, as TechCrunch first reported. Faraday was able to independently reproduce findings from published scientific papers without prior knowledge of the answers.

Inherent reports that Faraday achieved these results while utilizing a significantly smaller model, Qwen 3.6, which contains 27 billion parameters. This performance was measured against much larger systems, specifically OpenAI’s GPT-5.5 and Anthropic’s Claude Opus 4.8. To achieve this, Inherent utilized reinforcement learning to instill what cofounder and chief scientist Edward Hughes describes as "research taste"—the ability to design and prioritize experiments—rather than relying on rules or the study of scientific conduct.

While the startup's ultimate goal is to build an AI capable of discovering new scientific knowledge, Hughes told TechCrunch that replicating papers serves as a standard training exercise for human scientists. Notably, Faraday does not handle all its own tasks; Inherent reports that the agent uses OpenAI’s GPT-5.5 Codex as its coding tool.

Founded by Hughes and three other cofounders—Louis Kirsch, Kaloyan Aleksiev, and Tantum Collins—the company recently emerged from stealth with a $50 million seed round. Based in King's Cross, London, the firm currently employs 12 people and plans to expand its headcount to between 20 and 25 by the end of the year.