Kimi Models — Hardware Requirements
5 Kimi models from Moonshot AI and the community, from the smallest that runs in 10.7 GB of VRAM at Q4 up to 72.7B parameters. Every row links to full quantization tables and GPU compatibility.
All Kimi Models by Size
| Model | Params | Runs from (Q4) | Context | Publisher | Quant downloads |
|---|---|---|---|---|---|
| Kimi VL A3B Instruct | 16.4B | 10.7 GB | 131K | ||
| Kimi Linear 48B A3B Instruct | 49.1B | 33.9 GB | — | ||
| Kimi Linear 48B A3B Base | 49.1B | 33.9 GB | — |
Community variants and fine-tunes(1)
| Model | Params | Runs from (Q4) | Context | Publisher | Quant downloads |
|---|---|---|---|---|---|
| Kimi Dev 72B | 72.7B | 49.3 GB | 131K |
Frequently Asked Questions
- How much VRAM do I need to run a Kimi model?
- The smallest Kimi model, Kimi VL A3B Instruct, runs from 10.7 GB of VRAM at Q4_K_M (smaller low-bit quantizations start at 5.3 GB with a noticeable quality loss). Larger family members need proportionally more — see the table above for every model.
- Which Kimi models can I run on a 16 GB GPU?
- 1 of 4 Kimi models fit in 16 GB of VRAM at Q4_K_M, including Kimi VL A3B Instruct.
- What is the most popular Kimi model to run locally?
- Kimi Linear 48B A3B Instruct is the most downloaded Kimi model in local-friendly quantized formats. It runs from 33.9 GB of VRAM at Q4_K_M.