← Back to community benchmarks
Qwen3.6-27B
Performance
8k
tokens
333.1
PP tok/s
36.0
TG tok/s
24590
TTFT (ms)
17.3
Peak mem (GB)
Hardware
Chip
M3 Ultra (60c)
Memory
256 GB
GPU Cores
60
Software
oMLX
v0.5.2.dev1
macOS
macOS 27.0
Context
8,192
Performance by Context Length
| Context | PP tok/s | TG tok/s | Peak Mem | |
|---|---|---|---|---|
| 1k | 323.3 | 37.5 | 15.8 GB | view |
| 4k | 333.2 | 37.0 | 17.0 GB | view |
| 8k | 333.1 | 36.0 | 17.3 GB | current |
| 16k | 322.0 | 34.2 | 17.9 GB | view |
| 32k | 299.2 | 31.4 | 19.0 GB | view |
| 64k | 262.0 | 27.4 | 21.5 GB | view |
| 128k | 209.8 | 22.0 | 26.4 GB | view |
| 195k | 173.2 | 18.2 | 31.4 GB | view |
Batching Results
| Batch Size | TG tok/s | Speedup |
|---|---|---|
| 1× | 37.5 | 1.00× |
| 2× | 66.9 | 1.78× |