Kimi K3 uses fixed-size KDA state instead of a growing KV cache
vllm_project · x · 2026-07-27
- The post explains why fitting 2.8T parameters was not the hardest part for Kimi K3; caching them was.
- Many layers use a fixed-size KDA state instead of a growing KV cache, so there is no per-token KV to hash.
- Prefix caching was rebuilt around that design, and the author says every hybrid model after K3 can inherit the result.
- The attached diagram shows the model’s internal stack, including Kimi Delta Attention, Gated MLA, Stable LatentMoE, and the repeated block structure around KDA.
Related event: vLLM brings day-0 support to Moonshot’s Kimi K3(11 posts)→
More from Infra
- Together AI adds canary rollouts for zero-downtime model upgrades on dedicated inference — togethercompute · 2026-09-23
- Dedicated Hardware for Running AI Agents at Scale Arrives — cyrilzakka · 2026-09-23
- Ternary Bonsai 2 27B: 5.9GB weights retain ~95% of full-precision reasoning — cephaloform · 2026-09-23
- Qwen 27B runs 24hr unattended on one RTX5090, builds full Postgres-SpringBoot-React spreadsheet app — anglepoiselife · 2026-09-23
- OpenRoboto Shift launches: decentralized egocentric video data network for robot brains — markjeffrey · 2026-09-23
- Engineer describes designing digital circuits that recycle most of their energy — MikePFrank · 2026-09-23