Goodfire lands $150M Series B to make AI less of a black box
San Francisco-based AI research lab Goodfire has raised $150 million in a Series B funding round, valuing the company at $1.25 billion. The round was led by B Capital, with participation from existing backers Menlo Ventures, Lightspeed Venture Partners, South Park Commons, and Wing Venture Capital. New investors include DFJ Growth, Salesforce Ventures, and former Google CEO Eric Schmidt.
Building a “model design environment”
Goodfire says the new capital will accelerate development of what it calls a model design environment—a platform intended to help developers understand, debug, and deliberately shape AI system behavior at scale. The company argues that many modern models remain difficult to control because teams can observe outputs but lack clear visibility into the internal mechanisms that produce them.
Targeted edits instead of full retraining
Rather than retraining models from scratch, Goodfire is pursuing methods that allow researchers to reach into a model and adjust specific internal components tied to behavior. The company says it has demonstrated an approach that cut hallucinations in a large language model by nearly half by directly modifying internal mechanisms.
Research-first “neolabs” gain momentum
Led by CEO Eric Ho, Goodfire positions itself among a growing set of research-driven AI “neolabs” focused on interpretability and model understanding, rather than relying primarily on scaling compute and data. The team includes researchers with backgrounds at OpenAI and Google DeepMind, including interpretability contributors such as Nick Cammarata and co-founder Tom McGrath.
“Interpretability, for us, is the toolset for a new domain of science,” Ho said, describing a push toward designing intelligence more intentionally. The company also cited scientific applications, including work with partners such as the Mayo Clinic and the Arc Institute aimed at reverse-engineering models used in biomedical research.
Related
