SmolLM3 3B
Apache 2.0HuggingFace · 3B · Dense
Lightweight multilingual reasoning
830.0K downloads
992 likes
2025-07 128K context
Use Cases
chat reasoning
| Quant | Bits | VRAM | Quality | Status |
|---|---|---|---|---|
| Q2_K | 2 | 1.5 GB | low | — |
| Q3_K_M | 3 | 1.8 GB | moderate | — |
| Q4_K_M | 4 | 2 GB | good | — |
| Q5_K_M | 5 | 2.4 GB | good | — |
| Q6_K | 6 | 2.8 GB | excellent | — |
| Q8_0 | 8 | 3.6 GB | excellent | — |
| F16 | 16 | 6.6 GB | lossless | — |
GPUs that run SmolLM3 3B
SmolLM3 3B needs about 2 GB of VRAM at Q4_K_M, so it runs comfortably on a GPU with 3 GB of VRAM or more (193 of the GPUs we track). Common cards, smallest first:
| VRAM | GPUs |
|---|---|
| 3 GB | GTX 1060 3GB |
| 4 GB | RX 6500 XT , GTX 1650 SUPER , GTX 1650 Ti |
| 6 GB | RTX 2060 , Arc A380 , RX 5600 XT |
| 8 GB | RTX 5060 Ti , RX 7600 , Arc A750 |
About this model
Lightweight multilingual reasoning