← Back to community benchmarks
Qwen3-Coder-Next-mxfp4
Performance
32k
tokens
974.5
PP tok/s
58.3
TG tok/s
33625
TTFT (ms)
42.4
Peak mem (GB)
Hardware
Chip
M4 Max (40c)
Memory
128 GB
GPU Cores
40
Software
oMLX
v0.2.8
macOS
macOS 26.3.1
Context
32,768
Performance by Context Length
| Context | PP tok/s | TG tok/s | Peak Mem | |
|---|---|---|---|---|
| 1k | 1,096 | 86.9 | 40.7 GB | view |
| 4k | 1,241 | 77.3 | 40.9 GB | view |
| 8k | 1,233 | 73.9 | 41.1 GB | view |
| 16k | 1,156 | 69.7 | 41.5 GB | view |
| 32k | 974.5 | 58.3 | 42.4 GB | current |
| 64k | 777.6 | 44.9 | 44.2 GB | view |
Batching Results
| Batch Size | TG tok/s | Speedup |
|---|---|---|
| 1× | 86.9 | 1.00× |
| 2× | 140.6 | 1.62× |
| 4× | 204.8 | 2.36× |
| 8× | 285.2 | 3.28× |