Yes — with 2.3 GB to spare

Gemma 3 12B at Q4_K_M fits your GeForce RTX 3080 Ti entirely on the GPU at 8K context, at an estimated 80 tokens per second. Past 45K the KV cache pushes it over — quantise the cache to q8_0, or step down a quantisation, to go longer.

Fully 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.

What hardware do I need for Gemma 3 12B? →

The VRAM budget

weights 6.9 GB

Weights 6.9 GB KV cache @ 8K 0.81 GB Runtime overhead 0.6 GB Free 2.3 GB of 10.6 GB

Quantisation ladder

Quant	Weights	Total @ 8K	Max context	Tok/s	Quality	Fit
F16	22.7 GB	24.1 GB	—	~2.6	Reference	13.5 GB over
Q8_0	12.1 GB	13.5 GB	—	~10	−0.1% ppl	2.9 GB over
Q6_K	9.3 GB	10.7 GB	5K	~49	−0.4% ppl	0.1 GB over
Q5_K_M	8.1 GB	9.5 GB	26K	69	−0.8% ppl	Long context
Q4_K_M	6.9 GB	8.3 GB	45K	80	−1.9% ppl	Recommended
Q3_K_M	5.6 GB	7.0 GB	66K	99	−5.4% ppl	Long context
Q2_K	4.8 GB	6.2 GB	78K	116	−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

Ollama llama.cpp LM Studio vLLM

terminal

$ ollama pull gemma3:12b
$ OLLAMA_CONTEXT_LENGTH=8192 \
    ollama run gemma3:12b

The default. One binary, a model registry, an OpenAI-compatible port. More on Ollama.

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.

03There is room to go to 45K context on this card.

See all models for this rig Compare with another model