← Back to community benchmarks
Qwen3.6-27B-oQ6-mtp
Performance
32k
tokens
374.9
PP tok/s
31.2
TG tok/s
87396
TTFT (ms)
29.9
Peak mem (GB)
Hardware
Chip
M3 Ultra (80c)
Memory
512 GB
GPU Cores
80
Software
oMLX
v0.3.9.dev1
macOS
macOS 26.2
Context
32,768
Performance by Context Length
| Context | PP tok/s | TG tok/s | Peak Mem | |
|---|---|---|---|---|
| 1k | 381.3 | 35.2 | 23.0 GB | view |
| 4k | 411.3 | 37.1 | 24.4 GB | view |
| 8k | 406.6 | 34.6 | 25.4 GB | view |
| 16k | 395.7 | 32.9 | 26.9 GB | view |
| 32k | 374.9 | 31.2 | 29.9 GB | current |
Batching Results
| Batch Size | TG tok/s | Speedup |
|---|---|---|
| 1× | 35.2 | 1.00× |
| 2× | 45.1 | 1.28× |
| 4× | 56.0 | 1.59× |
| 8× | 62.5 | 1.78× |