Written by Jakub Rusinowski · Last updated November 20, 2024
Yes — comfortably
Yes, comfortably — SmolLM2 1.7B Instruct at Q8_0 needs about 4.2 GB of the 16 GB usable on RTX 5080 Desktop (16 GB VRAM, 32 GB RAM), leaving ~11.8 GB spare and running at ~169.8 tok/s (estimated), with room for about 8,192 tokens of context.
Confidence: high · Recommended quantization: Q8_0 · Estimated speed: ~169.8 tok/s
| Usable memory for models | 16 GB |
| Memory bandwidth | 960 GB/s |
| Form factor | Desktop |
| Operating system | Windows or Linux |
| Memory upgradeable | Yes |
| Quant | Memory needed | Fits 16 GB? | Max context | Est. speed | Download |
|---|---|---|---|---|---|
| F16 | 5.8 GB | ✓ Yes | 8K | ~123.2 tok/s | 3.4 GB |
| Q8_0 | 4.2 GB | ✓ Yes | 8K | ~169.8 tok/s | 1.8 GB |
| Q6_K | 3.8 GB | ✓ Yes | 8K | ~188.3 tok/s | 1.4 GB |
| Q5_K_M | 3.6 GB | ✓ Yes | 8K | ~198.1 tok/s | 1.2 GB |
| Q4_K_M | 3.4 GB | ✓ Yes | 8K | ~208.4 tok/s | 1 GB |
| Q3_K_M | 3.1 GB | ✓ Yes | 8K | ~228.5 tok/s | 0.7 GB |
| Q2_K | 3 GB | ✓ Yes | 8K | ~241.2 tok/s | 0.6 GB |
| Variant | Needs at Q4_K_M | Fits? | Est. speed |
|---|---|---|---|
| SmolLM2 1.7B Instruct | 3.4 GB | ✓ Fits | ~208.4 tok/s |
| SmolLM2 360M Instruct | 1.4 GB | ✓ Fits | ~359.4 tok/s |
Ollama or llama.cpp (CUDA/ROCm) — vLLM if you need concurrent serving
Yes, comfortably — SmolLM2 1.7B Instruct at Q8_0 needs about 4.2 GB of the 16 GB usable on RTX 5080 Desktop (16 GB VRAM, 32 GB RAM), leaving ~11.8 GB spare and running at ~169.8 tok/s (estimated), with room for about 8,192 tokens of context.
Q8_0 — it needs about 4.2 GB of the 16 GB available, downloads as roughly 1.8 GB, and runs at an estimated 169.8 tokens/sec with up to 8K of context.
Nothing binding — the model fits with headroom and generates at a usable speed on this hardware.
Ollama or llama.cpp (CUDA/ROCm) — vLLM if you need concurrent serving