← Back to community benchmarks

DeepSeek-R1-Distill-Qwen-14B

M4 (10c) · 24 GB · 4bit · 2026-05-18

Performance

1k

tokens

105.6

PP tok/s

11.2

TG tok/s

9702

TTFT (ms)

8.4

Peak mem (GB)

Hardware

Chip M4 (10c)

Memory 24 GB

GPU Cores 10

Software

oMLX v0.3.9.dev2

macOS macOS 26.5

Context 1,024

Performance by Context Length

Context	PP tok/s	TG tok/s	Peak Mem
1k	105.6	11.2	8.4 GB	current
4k	97.1	10.5	9.2 GB	view

Batching Results

Batch Size	TG tok/s	Speedup
1×	11.2	1.00×
2×	22.3	1.99×
4×	21.6	1.93×