Only with CPU offload

Mistral Small 3.2 24B at Q4_K_M needs 15.1 GB but only 7.0 GB is addressable, so about 61% of the layers would stream from system RAM at roughly 60 GB/s. Expect around 4.2 tokens per second — usable for batch work, painful for chat.

39% on GPU 8K context Q4_K_M · 13.3 GB Apache 2.0 Released Jun 2025 Vision

Apache-2.0, vision-capable, and the most 24 GB-friendly of the 2025 generalists.

Hardware requirements What hardware do I need to run Mistral Small 3.2 24B? →

At Q4_K_M with an 8K context: a 20 GB GPU (RTX 4000 Ada) or a 24 GB Mac (M4 Pro). Minimum, recommended, high-quality and long-context tiers, with all 114 devices checked at every quantisation.

The VRAM budget

weights 13.3 GB
Weights 13.3 GB KV cache @ 8K 1.25 GB Runtime overhead 0.6 GB Over budget 8.1 GB past 7.0 GB

Quantisation ladder

QuantWeightsTotal @ 8KMax contextTok/sQualityFit
Q8_0 23.4 GB 25.2 GB — ~1.9 −0.1% ppl 18.2 GB over
Q6_K 18.0 GB 19.9 GB — ~2.7 −0.4% ppl 12.9 GB over
Q5_K_M 15.6 GB 17.4 GB — ~3.3 −0.8% ppl 10.4 GB over
Q4_K_M 13.3 GB 15.1 GB — ~4.2 −1.9% ppl 8.1 GB over
Q3_K_M 10.7 GB 12.6 GB — ~5.8 −5.4% ppl 5.6 GB over
Q2_K 9.2 GB 11.1 GB — ~7.8 −15% ppl 4.1 GB over

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-Small-3.2-24B-Instruct-2506:Q4_K_M \
    -c 8192 -ngl 15

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

01Download is 13.3 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.
Mistral Small 4 119B-A6B
119B MoE · same family
Q4_K_M · 67.7 GB ~4.2 CPU offload
Devstral Small 2 24B
24B · similar size
Q4_K_M · 15.3 GB ~4.1 CPU offload
Gemma 4 26B-A4B
26.5B MoE · similar size
Q4_K_M · 16.0 GB ~10 CPU offload
gpt-oss 20B
20.9B MoE · similar size
MXFP4 · 11.6 GB ~15 CPU offload
Gemma 3 27B
27.4B · similar size
Q4_K_M · 17.0 GB ~3.4 CPU offload
Qwen3.8 27B
27.8B · similar size
Q4_K_M · 16.7 GB ~3.5 CPU offload
Qwen3.6 27B
27.8B · similar size
Q4_K_M · 16.7 GB ~3.5 CPU offload

Every model against the GeForce RTX 2080 Super →

Mistral Small 3.2 24B on other hardware

DeviceTotalTok/s est.
GeForce RTX 4060 Laptop
7.0 GB usable · 256 GB/s
15.1 GB ~3.9 CPU offload
GeForce RTX 3060 12 GB
10.6 GB usable · 360 GB/s
15.1 GB ~6.1 CPU offload
M4 · 16 GB
10.7 GB usable · 120 GB/s
15.1 GB — Won't fit
RTX 4000 Ada
18.4 GB usable · 360 GB/s
15.1 GB 16 Runs great
GeForce RTX 4090
22.4 GB usable · 1008 GB/s
15.1 GB 46 Runs great
Radeon RX 7900 XTX
22.4 GB usable · 960 GB/s
15.1 GB 44 Runs great
M4 Pro · 24 GB
18.0 GB usable · 273 GB/s
15.1 GB 11 Runs great
M3 Max · 64 GB
48.0 GB usable · 400 GB/s
15.1 GB 17 Runs great

All 114 devices, by tier →