Yes — with 4.7 GB to spare
Ministral 3 14B at Q4_K_M fits your RTX 2000 Ada entirely on the GPU at 8K context, at an estimated 17 tokens per second. Past 38K 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 · 7.8 GB
Apache 2.0
Released Dec 2025
Vision
The largest Ministral. A 12 GB card runs it at Q4 with a few gigabytes to spare.
The VRAM budget
weights 7.8 GB
Weights 7.8 GB
KV cache @ 8K 1.25 GB
Runtime overhead 0.6 GB
Free 4.7 GB of 14.4 GB
Quantisation ladder
| Quant | Weights | Total @ 8K | Max context | Tok/s | Quality | Fit |
|---|---|---|---|---|---|---|
| F16 | 25.9 GB | 27.7 GB | — | ~2.2 | Reference | 13.3 GB over |
| Q8_0 | 13.8 GB | 15.6 GB | — | ~8.0 | −0.1% ppl | 1.2 GB over |
| Q6_K | 10.6 GB | 12.5 GB | 20K | 13 | −0.4% ppl | Long context |
| Q5_K_M | 9.2 GB | 11.0 GB | 29K | 15 | −0.8% ppl | Long context |
| Q4_K_M | 7.8 GB | 9.7 GB | 38K | 17 | −1.9% ppl | Recommended |
| Q3_K_M | 6.3 GB | 8.2 GB | 47K | 21 | −5.4% ppl | Long context |
| Q2_K | 5.4 GB | 7.3 GB | 53K | 25 | −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.
How to run it
$ ollama pull ministral-3:14b $ OLLAMA_CONTEXT_LENGTH=8192 \ ollama run ministral-3:14b
The default. One binary, a model registry, an OpenAI-compatible port. More on Ollama.
01Download is 7.8 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 38K context on this card.