← Back to community benchmarks

Qwen2.5-3B-Instruct

M1 Max (24c) · 64 GB · unknown · 2026-05-18

Performance

1k

tokens

822.4

PP tok/s

96.5

TG tok/s

1245

TTFT (ms)

2.5

Peak mem (GB)

Hardware

Chip M1 Max (24c)

Memory 64 GB

GPU Cores 24

Software

oMLX v0.3.9.dev2

macOS macOS 15.7.4

Context 1,024

Performance by Context Length

Context	PP tok/s	TG tok/s	Peak Mem
1k	822.4	96.5	2.5 GB	current
4k	819.7	76.7	2.6 GB	view

Batching Results

Batch Size	TG tok/s	Speedup
1×	96.5	1.00×
2×	112.2	1.16×
4×	128.1	1.33×