SmolLM 360M punches above its weight for such a tiny model — it handles basic text tasks, simple Q&A, and lightweight reasoning with surprising coherence. Its real personality is defined by constraint: at 360M parameters, it runs on edge devices and low-memory environments where larger models simply can't fit. Expect capable but bounded outputs; complex reasoning or nuanced generation will hit walls quickly.
| Benchmark | Score | Type | Recorded |
|---|---|---|---|
| IFEval | 21.3 | accuracy | 3d ago |