can it run
Yes, but close to the edge. 12GB against a 10GB working set: it runs, with little room for long context windows or other processes.
weights 8GB + KV cache 3.12GB at 4k context vs ~6GB usable → no. 11.1 GB vs 6 GB usable — does not fit
computed roofline: 336 GB/s × 0.48–0.72 efficiency window / 8.0GB (Q4_K_M) — bands, never points.
| model file | 8.0GB (GGUF Q4_K_M) |
| minimum memory | 10GB working set |
| this machine | 12GB GDDR6, 336 GB/s |
| params | 13B |
| license | apache-2.0 |
| price | $375 |
easy run: ollama pull llava:13b
runs with: ollama · llama.cpp · koboldcpp · lm studio
→ Tesla P100 (16GB, used) — cheapest machine that runs it ($135)
→ every model the RTX 2060 12GB (used) can run — full list
→ rtx-5060-ti-16 — the next step up from this machine's GPU