llamaperf

M5 Max 128GB vs Radeon Pro V620 32GB

For running local LLMs · 21 reports across 10 models

Side A
M5 Max 128GB
Vendor
apple
VRAM
128GB
Memory
Unified
Side B
Radeon Pro V620 32GB
Vendor
amd
VRAM
32GB
Memory
Discrete

Highest reported generation speed by model family

These maxima come from different community setups. Model size, quantization, GPU count, offloading and concurrency may differ. They do not establish which GPU is faster for the same workload. Check the comparison conditions →

ModelM5 Max 128GBRadeon Pro V620 32GB
Qwen3.8up to 2400B55.0n=570.0n=4
Qwen3.6up to 35B79.4n=3
DeepSeek V4 Flashup to 285Bn=121.1n=1
GLM-5.3up to 321B24.0n=1
Gemma 4up to 31B7.5n=1
Museup to 30B35.4n=1
Nex-N2.5-mini133.6n=1
Qwen3.5up to 122Bn=1
StepFun 3.733.9n=1
Tencent-HY3up to 295B32.4n=1

More comparisons