MiniCPM Models — Hardware Requirements
18 MiniCPM models from OpenBMB and the community, from the smallest that runs in 0.5 GB of VRAM at Q4 up to 35.2B parameters. Every row links to full quantization tables and GPU compatibility.
All MiniCPM Models by Size
| Model | Params | Runs from (Q4) | Context | Publisher | Quant downloads |
|---|---|---|---|---|---|
| MiniCPM V 4.7 1B DSpark | 214M | 0.5 GB | 262K | ||
| MiniCPM5 2B DSpark | 324M | 0.6 GB | 131K | ||
| MiniCPM5 1B | 1.1B | 1.3 GB | 131K | ||
| MiniCPM5 1B Base | 1.1B | 1.3 GB | 131K | ||
| MiniCPM5 1B SFT | 1.1B | 1.3 GB | 131K | ||
| MiniCPM V 4.6 | 1.3B | 1.5 GB | 262K | ||
| MiniCPM V 4.6 Thinking | 1.3B | 1.5 GB | 262K | ||
| MiniCPM OCR 1.0 | 1.3B | 1.5 GB | 262K | ||
| MiniCPM V 4.7 1B | 1.6B | 1.6 GB | 262K | ||
| MiniCPM 2B Sft BF16 | 2B | 3 GB | 4K | ||
| MiniCPM5 2B | 2.5B | 2.5 GB | 131K | ||
| MiniCPM5 2B Base | 2.5B | 2.5 GB | 524K | ||
| MiniCPM V 4 | 4.1B | 3.1 GB | 33K | ||
| MiniCPM V 2 6 | 8.1B | 7.9 GB | — | ||
| MiniCPM O 2 6 | 8.7B | 6.4 GB | 33K | ||
| MiniCPM V 4 5 | 8.7B | 7.9 GB | 41K | ||
| MiniCPM O 4 5 | 9.4B | 8.3 GB | 41K | ||
| MiniCPM V 4.7 35B A3B DSpark | 35B | 21.5 GB | 262K | ||
| MiniCPM V 4.7 35B A3B | 35.2B | 22.1 GB | 262K |
Community variants and fine-tunes(9)
| Model | Params | Runs from (Q4) | Context | Publisher | Quant downloads |
|---|---|---|---|---|---|
| MiniCPM5 1B CoreAI | 1B | 2.1 GB | — | ||
| MiniCPM5 1B Agentic Tooluse Merged FP16 | 1.1B | 1.3 GB | 131K | ||
| Huihui MiniCPM5 1B Abliterated | 1.1B | 1.3 GB | 131K | ||
| MiniCPM5 1B Claude Opus Fable5 v2 Thinking Heretic | 1.1B | 1.3 GB | 131K | ||
| MiniCPM5 2B SFT | 2.5B | 2.5 GB | 131K | ||
| MiniCPM5 2B Midtrain | 2.5B | 2.5 GB | 131K | ||
| Minicpm5 2B Distilled Reasoning | 2.5B | 2.5 GB | 131K | ||
| MiniCPM SALA | 9.5B | 8.9 GB | 524K | ||
| MiniCPM MoE 8x2B | 8x2B | 11.4 GB | 4K |
Frequently Asked Questions
- How much VRAM do I need to run a MiniCPM model?
- The smallest MiniCPM model, MiniCPM V 4.7 1B DSpark, runs from 0.5 GB of VRAM at Q4_K_M (smaller low-bit quantizations start at 0.4 GB with a noticeable quality loss). Larger family members need proportionally more — see the table above for every model.
- Which MiniCPM models can I run on a 16 GB GPU?
- 26 of 28 MiniCPM models fit in 16 GB of VRAM at Q4_K_M, including MiniCPM SALA, MiniCPM V 4.7 1B, MiniCPM O 4 5.
- What is the most popular MiniCPM model to run locally?
- MiniCPM SALA is the most downloaded MiniCPM model in local-friendly quantized formats. It runs from 8.9 GB of VRAM at Q4_K_M.