llmgauge

run it locally

Run Mixtral 8x7B on RTX 4090

46.7B parameters on a 24 GB card. It won't fit in this card's memory at any usable quantization — here's the math, and what will run it.

verdict · Mixtral 8x7B on RTX 4090
Won't fit
even Q3_K_M needs 24.1 GB · only 22.8 GB usable
too big for this card — run it in the cloud
Smallest GPU that fits at Q4_K_M: A100 40GB
Rent A100 40GB by the hour
VRAM needed and fit for Mixtral 8x7B on RTX 4090 by quantization.
quantization vram needed fits 24gb? tokens/sec quality
FP16 (full) 112.1 GB ✗ no — Reference quality — no quantization loss.
Q8_0 59.4 GB ✗ no — Near-lossless; rarely worth the extra space over Q6.
Q6_K 46.0 GB ✗ no — Virtually indistinguishable from full precision.
Q5_K_M 38.7 GB ✗ no — Minor loss; an excellent quality-vs-size balance.
Q4_K_M 31.4 GB ✗ no — Small but measurable loss; the popular default.
Q3_K_M 24.1 GB ✗ no — Noticeable degradation; only when you're tight on VRAM.

Weights = params × bytes/weight, +20% for KV cache & runtime; usable VRAM is 95% of nameplate. Tokens/sec is a bandwidth ceiling (1008 GB/s) — real throughput is lower with long context. Try other combinations →

Why Mixtral 8x7B won't fit a RTX 4090

Mixtral 8x7B has 46.7 billion parameters. Even quantized hard to Q3_K_M, its weights plus KV-cache overhead come to roughly 24 GB, well past the RTX 4090's 23 GB of usable VRAM. Spilling the overflow to system RAM (CPU offload) works but can cut throughput tenfold. To run it at the popular Q4_K_M default you'd want a A100 40GB or larger — or rent one by the hour rather than buy.

Quantization is the lever

Each step down in precision shrinks the model: FP16 needs about 112 GB, Q4_K_M about 31 GB — a 72% reduction for a small, usually acceptable quality cost. Q5_K_M and Q6_K are near-lossless if you have the headroom; drop to Q3 only when you're genuinely out of VRAM. The quantization guide covers the tradeoffs in detail.

Frequently asked questions

Can a RTX 4090 run Mixtral 8x7B?

Not in VRAM. Even at the smallest practical quantization (Q3_K_M), Mixtral 8x7B needs about 24 GB versus the RTX 4090's 23 GB usable. You'd need a larger GPU — a A100 40GB fits it at Q4_K_M, or to rent one by the hour.

How much VRAM does Mixtral 8x7B need?

Mixtral 8x7B is a 46.7B-parameter model. At FP16 that's about 112 GB; at Q4_K_M (the popular default) about 31 GB, including ~20% for the KV cache and runtime. Quantization is the main lever — see the per-quant table above.

Other combinations

Mixtral 8x7B on other GPUs: RTX 3060 12GB, RTX 4070 Ti, RTX 4080, RTX 3090, NVIDIA L4, RTX A6000

Other models on the RTX 4090: Gemma 2 9B, Phi-3 Medium 14B, Gemma 2 27B, Qwen2.5 32B, Command R 35B, Llama 3.3 70B

Related