Yes, just — 0.1 GB spare

Mistral NeMo 12B at Q4_K_M fits your Arc B570 entirely on the GPU at 8K context, at an estimated 34 tokens per second. Past 8K 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 Apache 2.0 Released Jul 2024

Multilingual 12B with a 128K window, built with NVIDIA. A roleplay and fiction favourite that refuses to die.

Hardware requirements What hardware do I need to run Mistral NeMo 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.

The VRAM budget

weights 6.9 GB
Weights 6.9 GB KV cache @ 8K 1.25 GB Runtime overhead 0.6 GB Free 0.1 GB of 8.8 GB

Quantisation ladder

QuantWeightsTotal @ 8KMax contextTok/sQualityFit
F16 22.7 GB 24.6 GB — ~2.2 Reference 15.8 GB over
Q8_0 12.1 GB 13.9 GB — ~5.8 −0.1% ppl 5.1 GB over
Q6_K 9.3 GB 11.2 GB — ~10 −0.4% ppl 2.4 GB over
Q5_K_M 8.1 GB 9.9 GB — ~17 −0.8% ppl 1.1 GB over
Q4_K_M 6.9 GB 8.7 GB 8K 34 −1.9% ppl Recommended
Q3_K_M 5.6 GB 7.4 GB 16K 41 −5.4% ppl Fits
Q2_K 4.8 GB 6.6 GB 22K 48 −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

terminal
$ llama-server \
    -hf mistralai/Mistral-Nemo-Instruct-2407:Q4_K_M \
    -c 8192 -ngl 99

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.
03Only 0.1 GB is spare, so a long system prompt can still push it over. Its real ceiling here is 8K context.
See all models for this rig Compare with another model

Related models on an Arc B570

ModelTotalTok/s est.
Mistral 7B Instruct v0.3
7.25B · same family
Q4_K_M · 5.7 GB 56 Runs great
Gemma 3 12B
12.2B · similar size
Q4_K_M · 8.3 GB 34 Tight fit
Gemma 4 12B
12B · similar size
Q4_K_M · 8.2 GB 34 Tight fit
Ministral 3 14B
13.9B · similar size
Q4_K_M · 9.7 GB ~19 CPU offload
Phi-4 14B
14.7B · similar size
Q4_K_M · 10.4 GB ~14 CPU offload
Qwen3 14B
14.8B · similar size
Q4_K_M · 10.2 GB ~15 CPU offload
DeepSeek-R1-Distill-Qwen 14B
14.8B · similar size
Q4_K_M · 10.4 GB ~14 CPU offload

Every model against the Arc B570 →

Mistral NeMo 12B on other hardware

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

All 114 devices, by tier →