Kimi K3: When "Bigger" Doesn't Mean "Slower”

About this post

Moonshot AI just dropped Kimi K3 — an open-weight model with 2.8 trillion parameters, making it the world’s first open 3-trillion-parameter-class model. The twist? Only 104 billion of those parameters activate per token, thanks to a Mixture-of-Experts design routing through 16 of 896 experts. Add a 1-million-token context window and native vision, and you get a model built for long-horizon coding and reasoning — competitive with frontier closed models, but one anyone can self-host.