Databricks Tops Kimi K3 Inference at 239 tokens/s

Databricks has taken the top spot on Artificial Analysis for Kimi K3 inference speed and latency, achieving a blazing fast 239 tokens/s and setting a new industry benchmark.

2026-08-04 ~ 2026-08-04 · 2 related posts

• Databricks says Kimi K3 now runs at 239 tokens per second on its serving stack — Yuchenj_UW · 2026-08-04

1 near-duplicate retellings: altryne