Kimi K3 — Moonshot AI 的本地 AI 模型
作者： Jakub Rusinowski · 最后更新：
Moonshot AI's July 2026 flagship and its largest model to date — a 2.8-trillion-parameter Mixture-of-Experts that landed in the July 16–27 window. K3 roughly doubles the expert count of Kimi K2.5 (1T) while keeping active parameters modest (~48B per token), pushing Moonshot's lead on agentic coding, tool use, and long-horizon reasoning further ahead of K2.6. Natively multimodal with a 256K context. At this scale it is a data-center / API model — self-hosting needs well over a terabyte of memory even at Q4, so most users reach it through the Kimi API or the smaller K2.x community builds.
变体
Kimi K3 最小的变体在 Q4_K_M 下约需 1562 GB 显存——量化权重加框架开销，不含 KV 缓存。
| 模型 | Q4 下显存 | 显存 | 上下文 | 运行 |
|---|---|---|---|---|
| Kimi K3 2.8T → 2.8T/104B active | ~1691.3 GB | 1,000,000 | 云端托管标签 | |
| Kimi K3.1 Pro → 2.8T (104B active) | ~1562.1 GB | 1,048,576 | 云端托管标签 |
显存为 Q4_K_M 下的量化权重加开销，与 GPU 与显存检测器使用同一引擎计算。
如何在本地运行 Kimi K3
安装 Ollama，然后拉取标签。
No local Ollama tag. Ollama lists kimi-k3 only as a cloud-hosted ':cloud' tag, which runs on Ollama's servers, not your machine.
在上方选择一个尺寸，查看它自己的显存、速度估算和安装命令。
许可证
Weights are downloadable and commercial use is permitted, subject to the licence’s acceptable-use terms.
适用于: Kimi K3 2.8TWeights are downloadable and commercial use is permitted, subject to the licence’s acceptable-use terms.
适用于: Kimi K3.1 Pro我的 GPU 能运行 Kimi K3 吗？
Kimi K3 — 常见问题
How much VRAM does Kimi K3 need?
Kimi K3 needs about 1562 GB VRAM at Q4_K_M quantization for its smallest variant. Variants: Kimi K3 2.8T (1691 GB, Q4_K_M); Kimi K3.1 Pro (1562 GB, MXFP4). On Apple Silicon, unified memory counts toward this requirement.
Can I run Kimi K3 on an RTX 4090 (24 GB)?
Kimi K3's smallest variant needs about 1562 GB, which exceeds a single RTX 4090 (24 GB). Use multiple GPUs, a higher-VRAM card, or Apple Silicon with large unified memory.
What quantization should I use for Kimi K3?
Q4_K_M is the best balance of quality and VRAM for Kimi K3 in most cases. Choose Q8_0 for near-lossless quality if you have spare VRAM, or smaller quants (Q3/Q2) only when memory is tight.
How do I run Kimi K3 with Ollama?
Kimi K3 has no local Ollama tag — the published tag is cloud-hosted, so running it sends your prompts to a hosted GPU rather than your own machine.