VRAM CalculatorCan I Run
✅ Yes — fits in VRAM

Mistral Small 22BonRTX 4090

Affiliate disclosure: We earn commissions when you shop through the links below at no additional cost to you.

MHA model · 8 attention heads · 8k context · llama.cpp

14.5 GiB
+9.5 GiB headroom

estimated

Model
Mistral Small 22B
22B params · Dense · 36 layers
GPU
RTX 4090
24GB VRAM · NVIDIA
Enthusiast GPU

VRAM by Quantization

Calculated estimate using: Weights + KV Cache (GQA-aware) + Activations + Engine Overhead. These are theoretical calculations — actual VRAM usage varies by runtime and batch size.

FP16 (Full precision)2 bpw
43.9 GiB— OOM
Q8_0 (High quality)1 bpw
23.4 GiB— tight
Q5_K_M (Balanced)0.6875 bpw
17.0 GiB— fits
Q4_K_M (Efficient)0.5625 bpw
14.5 GiB— fits
IQ4_XS (Ultra efficient)0.53 bpw
13.8 GiB— fits

KV cache always stored in FP16 (2 bytes). GQA reduces KV cache size by 1.0× vs standard MHA.

Confidence

Medium

Standard inference workload estimated with llama.cpp engine overhead at 0.8GB. Variance depends on batch size, exact context used, and runtime implementation. This is a calculated estimate, not a measured runtime value.

Deploy a GPU API

Your GPU fits the model. Going cloud? Use managed GPU to avoid infra headaches.

Affiliate Disclosure

Some links on this page are affiliate links. If you purchase through them, we may earn a small commission at no additional cost to you. We only recommend tools we've thoroughly researched and believe add real value.

Our reviews and comparisons are based on objective analysis and are not influenced by affiliate partnerships.

Other combinations: