← Back to community benchmarks

DeepSeek-V4-Flash

M3 Ultra (80c) · 256 GB · 4bit · 2026-05-18

Performance

1k

tokens

381.8

PP tok/s

35.0

TG tok/s

2682

TTFT (ms)

142.2

Peak mem (GB)

Hardware

Chip M3 Ultra (80c)

Memory 256 GB

GPU Cores 80

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	381.8	35.0	142.2 GB	current
4k	440.2	32.1	142.4 GB	view

Batching Results

Batch Size	TG tok/s	Speedup
1×	35.0	1.00×
2×	53.7	1.53×
4×	78.8	2.25×