gpu4.pics
data via LocalScore
NVIDIA RTX A6000
48 GB VRAM NVIDIA 300 W TDP
running Meta Llama 3.1 8B Instruct· Q4_K_M
90.5
tokens / sec
938 · Strong #16 of 370
Generation
90.5tok/s
higher is better
Prompt / prefill
3,447tok/s
higher is better
Time to first token
378 ms
lower is better
Composite score
938
Model size
8B
Runtime
gpu4.pics/r/ls-nvidia-rtx-a6000-meta-llama-31-8b-instruct-q4-k-m

All benchmarks on this rig

ModelQuantGenPromptTTFTPowerPerf/W$/MtokScore
Meta Llama 3.1 8B InstructQ4_K_M90.53,447378 ms938

How it stacks up — Meta Llama 3.1 8B Instruct

NVIDIA RTX A6000
90.5
NVIDIA H100
120
NVIDIA A100 80GB
110
NVIDIA RTX 3090 Ti
110
NVIDIA RTX PRO 6000 Blackwell Workstation Edition
103
NVIDIA RTX 3090
103