SmolLM3
SmolLM3-3B
Hugging Face · open-permissive apache-2.0
| Parameters (B) | 3.075 |
|---|---|
| Architecture | dense |
| Layers / KV heads / head dim | 36 / 4 / 128 |
| Context window | 65536 |
| Modalities | in: text; out: text |
| GGUF | available |
| Commercial use | yes |
Minimum hardware (estimate)
At Q4_K_M and 8k context, total working set ≈ 3.0 GiB. On a machine with 64 GiB system RAM and no discrete GPU: CPU-only, slow. Limiting factor: No GPU VRAM; fits in system RAM but inference will be CPU-bound.
Scores
No Tier A comparable scores ingested yet for this model.
Provenance
- identity, license, architecture, parameter count: https://huggingface.co/HuggingFaceTB/SmolLM3-3B (retrieved 2026-08-10)
HF: HuggingFaceTB/SmolLM3-3B · snapshot 2026-08-10