← Back to community benchmarks

Llama-3.3-70B-Instruct

M5 Max (40c) · 128 GB · 6bit · 2026-04-04

Performance

1k

tokens

254.0

PP tok/s

7.9

TG tok/s

4036

TTFT (ms)

54.2

Peak mem (GB)

Hardware

Chip M5 Max (40c)

Memory 128 GB

GPU Cores 40

Software

oMLX v0.3.2

macOS macOS 26.4

Context 1,024

Performance by Context Length

Context	PP tok/s	TG tok/s	Peak Mem
1k	254.0	7.9	54.2 GB	current
4k	182.2	7.2	55.5 GB	view

Batching Results

Batch Size	TG tok/s	Speedup
1×	7.9	1.00×
2×	7.7	0.97×
4×	5.3	0.67×