GPUs / Apple

Apple M1

Memory
LPDDR4X
Bus
128-bit
Published peak
68.25 GB/s
Capacity
8 GB

Measured

Read
51.0GB/s
Write
52.5GB/s
Copy
52.1GB/s
Random
3.82GB/s
Latency
373ns
Sustain
51.5GB/s

Medians of 1 verified run from 1 operator. Read reaches 75 % of the published peak.

Local AI speed

ModelParameters4-bit, tokens/s8-bit, tokens/s
Llama 3.1 8B8.03B≤ 13Does not fit
Qwen 2.5 14B14.7B≤ 6.9Does not fit
gpt-oss-20b21B, 3.6B activeDoes not fitDoes not fit
Mistral Small 24B24BDoes not fitDoes not fit
Qwen 2.5 32B32.5BDoes not fitDoes not fit
Llama 3.3 70B70.6BDoes not fitDoes not fit
gpt-oss-120b117B, 5.1B activeDoes not fitDoes not fit
Upper bounds: median read ÷ bytes read per token. Mixture-of-experts models read only their active experts. Fit checks weights against the 8 GB configuration; the KV cache needs room on top.

Top operators

#OperatorScoreReadDate
1OP-8B53B834.051.0 GB/s2026-09-27

Full Apple M1 leaderboard