NASA runs Google's Gemma AI model in orbit to analyze satellite images

NASA's Jet Propulsion Laboratory deployed Google's Gemma 3 vision-language model on a YAM-9 satellite built by Loft Orbital, in what it describes as the first in-orbit demonstration of a vision-language model analyzing imagery from a satellite's own sensors. The system, called NAVI-Orbital, lets researchers send natural-language prompts to the spacecraft instead of writing structured operator commands.

Why it matters: This shows open-weight models are now compact and efficient enough to run directly on spacecraft hardware, potentially letting scientists interact with satellites via plain-language prompts rather than needing a dedicated operations team — a meaningful shift in how space missions could be run.