← Back to community benchmarks

Qwen3-Coder-Next

M3 Ultra (60c) · 256 GB · 4bit · 2026-03-08
Performance
16k
tokens
1,461
PP tok/s
58.9
TG tok/s
11213
TTFT (ms)
43.9
Peak mem (GB)
Hardware
Chip M3 Ultra (60c)
Memory 256 GB
GPU Cores 60
Software
oMLX v0.2.6
macOS macOS 26.3
Context 16,384
Performance by Context Length
Context PP tok/s TG tok/s Peak Mem
1k 1,298 73.9 43.0 GB view
4k 1,558 71.0 43.2 GB view
8k 1,552 66.4 43.4 GB view
16k 1,461 58.9 43.9 GB current
32k 1,281 49.4 44.7 GB view
64k 1,007 36.9 46.5 GB view
Batching Results
Batch Size TG tok/s Speedup
73.9 1.00×
125.4 1.70×
196.9 2.66×
286.3 3.87×