Baseten has launched Base Labs, a research arm, alongside a new safety infrastructure standard for open-weight models, in partnership with Hugging Face and Goodfire AI. The announcement comes amid debate over the safety of open-weight models, which can be weakened by abliteration, a technique that removes safeguards. Hugging Face currently lists over 6,000 abliterated models.

Base Labs will develop and publish methods for training and monitoring open models. The company frames the work as a transparent standard built into how models are trained and deployed, rather than added afterward. Baseten argues openness provides more visibility into model behavior and better means for turning safety research into actionable, transparent controls.

The companies have not disclosed the technical details of the partnership. Goodfire, which specializes in interpreting AI decision-making, is likely to contribute the “built into” part. Baseten, an AI inference provider, raised a $1.5 billion Series F in June, reaching a $13 billion valuation. Goodfire raised a $150 million Series B led by B Capital earlier this year.

Baseten has issued an open call to the developer ecosystem to contribute to the framework, aiming to build an ecosystem of open models that are safe and accessible.