Skip to content
llm-speed
← raw run
wrapped·Apr 28, 2026
/r/r_sqzp-0rdez-
your run

phi-4-4bit

on M3 Ultra
74.4
tok/sdecode
M3 Ultra (60-core GPU) + 96GB unified
rank in tier
12/21M3 Ultra runstop 57%
best workload
chat-short
where the rig flew
slowest workload
—
single-workload run
backend
mlxMLX is the fastest backend on Apple Silicon for dense models.
faster than
  • Qwen3.6-27B-Q4_K_M.gguf on RTX 509069.9 tok/s
  • <script>alert(1)</script> on Pentest-Bench42.0 tok/s
  • victim on Pentest-Bench42.0 tok/s

This is a shareable view of one signed llm-speed suite-suite-v1 run. Numbers are not edited. The raw record has every workload, the public key, and the fingerprint hash.