Vision4h ago
NAVI-Orbital Performs First In-Orbit VLM Inference
On April 16, 2026, NAVI-Orbital completed the first in-orbit demonstration of a vision-language model performing autonomous inference on a spacecraft. The system utilizes Gemma 3 to classify scenes and generate natural-language descriptions onboard. This shift reduces downlink bandwidth bottlenecks. Practitioners can now execute complex multi-modal queries without waiting for ground-station processing.