llamaperf
Calculator
GPUs
Engines
Guides
Blog
Community
MCP
Advertise
Submit
Sign in
Submit
llamaperf
/
M3 Max 96GB
/
report
DeepSeek V4 Flash 284B (13B active)
on
M3 Max 96GB
· ds4
Tone: positive
Apple hardware
Sep 7, 2026
Share
X
Reddit
LinkedIn
Bluesky
Copy
link
Throughput
12.7 t/s gen
System RAM
96 GB
Summary
User reports DeepSeek V4 Flash at 11-13 t/s on an M3 Max 96GB with SSD streaming and iogpu.wired_limit_mb=86016. Setup is antirez's ds4 engine with GGUF. TTFT is 3-5s after warmup. Prefill of 36k tokens takes about 2.5 minutes.
View source ↗
Feedback