Tweet by michellechen

July 8, 2026

big day for the homies. super excited to work with @vikhyatk and @jason_allen to bring the long awaited Moondream 3.1 to Workers AI. workers ai runs GPUs on cloudflare's edge, which is a natural infrastructure fit for real-time use cases like vision. moondream requests return in 160ms-700ms e2e. so fun to work with a team i've admired for so long. they even tuned specific kernels to fit our latency-tier GPUs. and we run this within a cog container, where we can leverage moondream's high-performance inference engine: photon. technical blog to come soon!

Author
michellechen
Date
July 8, 2026