vLLM previews production-scale support for Moonshot AI's Kimi K3 ahead of its weight release

vLLM / Moonshot AI

Tools official 1 src. ~1 min

The vLLM project published a July 22, 2026 blog post previewing production-scale support for Kimi K3, Moonshot AI's 2.8-trillion-parameter model with native vision and a 1-million-token context window, ahead of its scheduled July 27 weight release. Work includes separating physical KDA state-block size from prefix-match granularity for more efficient cache reuse, plus kernel, multimodal, and hardware-specific (NVIDIA and AMD) optimizations.

Why it matters

Shows inference engines racing to have day-one support ready for a massive open-weight model release, and previews new caching techniques for KDA-style attention variants.

Importance: 2/5

Notable inference-engine update ahead of a major open-weight model release.

Sources