Yes — with 38.1 GB to spare
Gemma 3 12B at Q4_K_M fits your RTX 6000 Ada entirely on the GPU at 8K context, at an estimated 85 tokens per second. There is room for its full 128K window.
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.
The VRAM budget
weights 6.9 GB
Weights 6.9 GB
KV cache @ 8K 0.81 GB
Runtime overhead 0.6 GB
Free 38.1 GB of 46.4 GB
Quantisation ladder
| Quant | Weights | Total @ 8K | Max context | Tok/s | Quality | Fit |
|---|---|---|---|---|---|---|
| F16 | 22.7 GB | 24.1 GB | 128K | 26 | Reference | Long context |
| Q8_0 | 12.1 GB | 13.5 GB | 128K | 48 | −0.1% ppl | Long context |
| Q6_K | 9.3 GB | 10.7 GB | 128K | 62 | −0.4% ppl | Long context |
| Q5_K_M | 8.1 GB | 9.5 GB | 128K | 72 | −0.8% ppl | Long context |
| Q4_K_M | 6.9 GB | 8.3 GB | 128K | 85 | −1.9% ppl | Recommended |
| Q3_K_M | 5.6 GB | 7.0 GB | 128K | 105 | −5.4% ppl | Long context |
| Q2_K | 4.8 GB | 6.2 GB | 128K | 122 | −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 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 the model's full 128K context on this card.