← Back to community benchmarks

Qwen3-Coder-Next

M3 Ultra (80c) · 256 GB · 8bit · 2026-03-13
Performance
8k
tokens
1,939
PP tok/s
60.3
TG tok/s
4226
TTFT (ms)
80.5
Peak mem (GB)
Hardware
Chip M3 Ultra (80c)
Memory 256 GB
GPU Cores 80
Software
oMLX v0.2.11
macOS macOS 26.3
Context 8,192
Performance by Context Length
Context PP tok/s TG tok/s Peak Mem
1k 1,572 64.9 80.1 GB view
4k 1,924 62.7 80.3 GB view
8k 1,939 60.3 80.5 GB current
16k 1,838 48.9 81.0 GB view
32k 1,609 26.9 81.8 GB view
64k 1,241 14.1 83.6 GB view
Batching Results
Batch Size TG tok/s Speedup
64.9 1.00×
111.9 1.72×
185.9 2.86×