On Kimi K3, Vera Rubin serves ~2.8x GB300's throughput per MW at ~67 tok/s per user, and sets the fastest Kimi K3 result on our leaderboard at 255 tok/s per user
Vera Rubin serves Kimi K3 at 255 tok/s per user, fastest on leaderboard
AIArtificial Analysis reports that Nvidia's Vera Rubin serves Kimi K3 at about 2.8x the throughput per megawatt of GB300 at roughly 67 tok/s per user. The same post says Vera Rubin sets the fastest Kimi K3 result on the Artificial Analysis leaderboard at 255 tok/s per user.
Source details
Artificial Analysis · x.comPublished
Original title: On Kimi K3, Vera Rubin serves ~2.8x GB300's throughput per MW at ~67 tok/s per user, and sets the fastest Kimi K3 result on our leaderboa...
