← Back to community benchmarks
Qwen3.6-27B
Performance
8k
tokens
653.4
PP tok/s
29.8
TG tok/s
12537
TTFT (ms)
18.3
Peak mem (GB)
Hardware
Chip
M5 Max (40c)
Memory
128 GB
GPU Cores
40
Software
oMLX
v0.5.1
macOS
macOS 26.5
Context
8,192
Performance by Context Length
| Context | PP tok/s | TG tok/s | Peak Mem | |
|---|---|---|---|---|
| 1k | 640.6 | 29.3 | 15.8 GB | view |
| 4k | 608.2 | 30.7 | 17.3 GB | view |
| 8k | 653.4 | 29.8 | 18.3 GB | current |
| 16k | 613.0 | 27.1 | 20.4 GB | view |
| 32k | 592.7 | 26.0 | 23.3 GB | view |
| 64k | 475.1 | 21.2 | 29.9 GB | view |
| 128k | 336.3 | 16.5 | 43.6 GB | view |
| 195k | 250.5 | 11.8 | 58.6 GB | view |
Batching Results
| Batch Size | TG tok/s | Speedup |
|---|---|---|
| 1× | 29.3 | 1.00× |
| 2× | 41.3 | 1.41× |
| 4× | 64.5 | 2.20× |