Only with CPU offload

Gemma 3 12B at Q4_K_M needs 8.3 GB but only 7.0 GB is addressable, so about 19% of the layers would stream from system RAM at roughly 60 GB/s. Expect around 19 tokens per second — usable for batch work, painful for chat.

81% on GPU 8K context Q4_K_M · 6.9 GB Gemma Terms of Use Released Mar 2025 Vision

Strong multilingual chat with images, sized for 12–16 GB cards. Gemma 4 12B is the same size and better.

Hardware requirements What hardware do I need to run Gemma 3 12B? →

At Q4_K_M with an 8K context: a 12 GB GPU (RTX 3060 12 GB) or a 16 GB Mac (M4). Minimum, recommended, high-quality and long-context tiers, with all 114 devices checked at every quantisation.

Fits instead: Gemma 3 4B (3.3 GB) · Gemma 3 1B (1.2 GB)

The VRAM budget

weights 6.9 GB
Weights 6.9 GB KV cache @ 8K 0.81 GB Runtime overhead 0.6 GB Over budget 1.3 GB past 7.0 GB

Quantisation ladder

QuantWeightsTotal @ 8KMax contextTok/sQualityFit
F16 22.7 GB 24.1 GB — ~2.0 Reference 17.1 GB over
Q8_0 12.1 GB 13.5 GB — ~5.1 −0.1% ppl 6.5 GB over
Q6_K 9.3 GB 10.7 GB — ~8.2 −0.4% ppl 3.7 GB over
Q5_K_M 8.1 GB 9.5 GB — ~12 −0.8% ppl 2.5 GB over
Q4_K_M 6.9 GB 8.3 GB — ~19 −1.9% ppl 1.3 GB over
Q3_K_M 5.6 GB 7.0 GB 8K 54 −5.4% ppl Fits
Q2_K 4.8 GB 6.2 GB 21K 63 −15% ppl Long context

Quality is the published perplexity delta against f16 weights. Max context assumes an f16 KV cache; q8_0 roughly doubles it. This model interleaves sliding-window layers (1024 tokens, 1 global in 6), which is why its cache barely grows with context.

How to run it

terminal
$ llama-server \
    -hf google/gemma-3-12b-it:Q4_K_M \
    -c 8192 -ngl 39

The engine underneath most of the others. Every knob is exposed. More on llama.cpp.

01Download is 6.9 GB. Keep it on an SSD — a first load off a spinning disk takes minutes.
02Close anything else holding VRAM. A browser with hardware acceleration can sit on 1–2 GB.
03If it falls back to CPU silently, drop the context first, then step down a quantisation.
See all models for this rig Compare with another model

Related models on a GeForce RTX 2080 Super

ModelTotalTok/s est.
Gemma 3 27B
27.4B · same family
Q4_K_M · 17.0 GB ~3.4 CPU offload
Gemma 3 4B
4.3B · same family
Q4_K_M · 3.3 GB 124 Runs great
Gemma 3 1B
1.0B · same family
Q4_K_M · 1.2 GB 534 Runs great
Mistral NeMo 12B
12.2B · similar size
Q4_K_M · 8.7 GB ~16 CPU offload
Gemma 4 12B
12B · similar size
Q4_K_M · 8.2 GB ~20 CPU offload
Ministral 3 14B
13.9B · similar size
Q4_K_M · 9.7 GB ~11 CPU offload
Phi-4 14B
14.7B · similar size
Q4_K_M · 10.4 GB ~9.1 CPU offload

Every model against the GeForce RTX 2080 Super →

Gemma 3 12B on other hardware

DeviceTotalTok/s est.
GeForce RTX 4060 Laptop
7.0 GB usable · 256 GB/s
8.3 GB ~14 CPU offload
GeForce RTX 3060 12 GB
10.6 GB usable · 360 GB/s
8.3 GB 32 Runs great
M4 · 16 GB
10.7 GB usable · 120 GB/s
8.3 GB 9.8 Runs great
GeForce RTX 4090
22.4 GB usable · 1008 GB/s
8.3 GB 89 Runs great
Radeon RX 7900 XTX
22.4 GB usable · 960 GB/s
8.3 GB 85 Runs great
M3 Max · 64 GB
48.0 GB usable · 400 GB/s
8.3 GB 33 Runs great

All 114 devices, by tier →