can it run
Comfortably. 8GB of memory against a 4GB working set leaves headroom for context, the OS, and anything else you're running.
weights 1GB + KV cache 0.1GB at 4k context vs ~8GB usable → full. 1.1 GB needed of 8 GB usable — headroom for context
computed roofline: 448 GB/s × 0.48–0.72 efficiency window / 1.0GB (Q4_K_M) — bands, never points. Small models (under ~4.5B) run BELOW these bands — kernel-launch overhead keeps them from saturating bandwidth (measured against community medians, 09-06).
| model file | 1.0GB (GGUF Q4_K_M) |
| minimum memory | 4GB working set |
| this machine | 8GB GDDR6, 448-496 GB/s |
| params | 1B |
| license | gemma |
| price | $305 |
easy run: ollama pull gemma3:1b
runs with: ollama · llama.cpp · koboldcpp · lm studio
→ every model the RTX 2080 / 2080 Super (8GB, used) can run — full list
→ rtx-3080-10gb — the next step up from this machine's GPU