Moonshot AI · 2.8T (104B active) · Mixture of Experts
K3 post-trained with large-scale RL — 2.8T multimodal MoE, 104B active, sparse attention for its 1M context, stronger coding and agentic work Check if your GPU or Mac can run Kimi K3.1 Pro locally — 1553.4 GB min, 2589.1 GB recommended.
2026-111024K context
Mixture of Experts
Total experts: 896
Active experts: 16
Active params: 104.0B
| Quant | Bits | VRAM | Quality | Status |
|---|---|---|---|---|
| Q2_K | 2 | 890.5 GB | low | — |
| Q3_K_M | 3 | 1246.5 GB | moderate | — |
| Q4_K_M | 4 | 1424.5 GB | good | — |
| Q5_K_M | 5 | 1780.5 GB | good | — |
| Q6_K | 6 | 2136.5 GB | excellent | — |
| Q8_0 | 8 | 2848.5 GB | excellent | — |
| F16 | 16 | 5696.5 GB | lossless | — |
Can I run Kimi K3.1 Pro locally?
- Can I run Kimi K3.1 Pro locally?
- Kimi K3.1 Pro needs about 1553.4 GB of memory at a minimum and 2589.1 GB recommended. Open this page to grade it against your GPU or Mac, then run it with runai, Ollama or LM Studio.
- How much VRAM does Kimi K3.1 Pro need?
- At Q4_K_M, Kimi K3.1 Pro uses about 1424.5 GB of VRAM. Higher quants need more memory; lower quants fit tighter cards with a quality tradeoff.