← Back to community benchmarks

Qwen3-Coder-Next

M4 Max (40c) · 128 GB · 6bit · 2026-04-07
Performance
8k
tokens
1,255
PP tok/s
63.6
TG tok/s
6526
TTFT (ms)
62.8
Peak mem (GB)
Hardware
Chip M4 Max (40c)
Memory 128 GB
GPU Cores 40
Software
oMLX v0.3.4
macOS macOS 26.3.1
Context 8,192
Performance by Context Length
Context PP tok/s TG tok/s Peak Mem
1k 1,074 68.1 61.6 GB view
4k 1,267 66.0 62.5 GB view
8k 1,255 63.6 62.8 GB current
16k 1,178 53.1 63.5 GB view
32k 1,026 39.4 64.9 GB view
Batching Results
Batch Size TG tok/s Speedup
68.1 1.00×
106.8 1.57×
148.2 2.18×
190.4 2.80×