can-i-run-this-llm

Can I run this LLM?

Can a NVIDIA RTX 3080 (10GB) run Gemma 4 12B?

Gemma · Apache 2.0 · estimated for a 10.0 GB machine at 4K context.

Yes, but it's tight.

Best fit: Gemma 4 12B at Q3_K_M · 5.1 GB · 60–89 tok/s on a NVIDIA RTX 3080 (10GB).

Every quantization on this machine

QuantSizeSpeedFit
Q3_K_M5.1 GB60–89 tok/sTight / slow
Q4_K_M7.2 GB49–73 tok/sWon't fit
Q5_K_M8.5 GB44–66 tok/sWon't fit
Q6_K9.8 GB40–60 tok/sWon't fit
Q8_012.7 GB33–50 tok/sWon't fit
FP1623.8 GB20–30 tok/sWon't fit

Ran Gemma 4 12B on a NVIDIA RTX 3080 (10GB)? Add your real numbers →