← Back to community benchmarks
Qwen36-27B-oQ8-mtp
Performance
16k
tokens
238.8
PP tok/s
23.9
TG tok/s
68611
TTFT (ms)
32.8
Peak mem (GB)
Hardware
Chip
M4 Max (40c)
Memory
128 GB
GPU Cores
40
Software
oMLX
v0.3.9.dev1
macOS
macOS 15.7.4
Context
16,384
Performance by Context Length
| Context | PP tok/s | TG tok/s | Peak Mem | |
|---|---|---|---|---|
| 1k | 235.7 | 27.0 | 28.8 GB | view |
| 4k | 246.5 | 26.6 | 30.3 GB | view |
| 8k | 244.8 | 24.9 | 31.3 GB | view |
| 16k | 238.8 | 23.9 | 32.8 GB | current |
| 32k | 225.7 | 22.9 | 35.8 GB | view |
| 64k | 200.9 | 19.4 | 41.8 GB | view |
Batching Results
| Batch Size | TG tok/s | Speedup |
|---|---|---|
| 1× | 27.0 | 1.00× |
| 2× | 32.0 | 1.19× |
| 4× | 45.6 | 1.69× |