Tag
Moondream's Photon inference engine eliminates GPU bubbles through pipelined decoding, achieving near-realtime VLM inference with up to 35% higher decode throughput.
Hermes Agent by Nous Research can now be used via iMessage through Photon's gateway setup.