RTX 5090~328 tok/s (Q4)

unsloth/Meta-Llama-3.1-8B-Instruct-bnb-4bit speed on RTX 5090

Quantization-specific throughput and VRAM requirements for unsloth/Meta-Llama-3.1-8B-Instruct-bnb-4bit running on RTX 5090.

Speed Snapshot

Topline estimate from compatibility data

Modelunsloth/Meta-Llama-3.1-8B-Instruct-bnb-4bit

GPURTX 5090

Q4 speed328 tok/s

Q4 VRAM required4GB

Data Source

Calculation and benchmark status

Speed values come from the compatibility dataset (`estimatedTokensPerSec`) and are sorted by quantization.

For full verdict logic and alternate GPUs, see the canonical compatibility page.

Quantization Speed Table

Quantization	VRAM needed	VRAM available	Speed	Verdict
Q4	4GB	32GB	328 tok/s	✅ Fits
Q8	9GB	32GB	228 tok/s	✅ Fits
FP16	17GB	32GB	103 tok/s	✅ Fits