← Back to community benchmarks

DeepSeek-R1-Distill-Qwen-32B

M3 Ultra (60c) · 96 GB · 4bit · 2026-05-18

Performance

1k

tokens

261.6

PP tok/s

32.8

TG tok/s

3929

TTFT (ms)

18.0

Peak mem (GB)

Hardware

Chip M3 Ultra (60c)

Memory 96 GB

GPU Cores 60

Software

oMLX v0.3.8

macOS macOS 26.5

Context 1,024

Performance by Context Length

Context	PP tok/s	TG tok/s	Peak Mem
1k	261.6	32.8	18.0 GB	current
4k	272.2	29.8	19.0 GB	view

Batching Results

Batch Size	TG tok/s	Speedup
1×	32.8	1.00×
2×	50.5	1.54×
4×	58.3	1.78×