← Back to community benchmarks
Qwen3-Coder-Next-5bit
Performance
32k
tokens
2,336
PP tok/s
53.9
TG tok/s
14028
TTFT (ms)
55.8
Peak mem (GB)
Hardware
Chip
M5 Max (40c)
Memory
128 GB
GPU Cores
40
Software
oMLX
v0.2.16
macOS
macOS 26.4
Context
32,768
Performance by Context Length
| Context | PP tok/s | TG tok/s | Peak Mem | |
|---|---|---|---|---|
| 1k | 1,911 | 86.2 | 52.3 GB | view |
| 4k | 2,847 | 80.5 | 53.3 GB | view |
| 8k | 2,983 | 77.4 | 53.7 GB | view |
| 16k | 2,761 | 73.8 | 54.4 GB | view |
| 32k | 2,336 | 53.9 | 55.8 GB | current |
| 64k | 1,727 | 34.9 | 58.6 GB | view |
| 128k | 1,040 | 20.7 | 64.2 GB | view |
| 195k | 706.6 | 13.6 | 70.1 GB | view |
Batching Results
| Batch Size | TG tok/s | Speedup |
|---|---|---|
| 1× | 86.2 | 1.00× |
| 2× | 138.6 | 1.61× |
| 4× | 184.9 | 2.15× |
| 8× | 227.4 | 2.64× |