llamaperf
Submit

M1 16GB vs RTX 3090

For running local LLMs · 7 reports across 2 models

Side A
M1 16GB
Vendor
apple
VRAM
16GB
Memory
Unified
Side B
RTX 3090
Vendor
nvidia
VRAM
24GB
Memory
Discrete

Tokens per second by model

ModelM1 16GBRTX 3090
Qwen3.6up to 35B—72.9n=6
Qwen2.5—28.0n=1

More comparisons