← Back to community benchmarks
Qwen3.6-27B-oQ4-fp16-mtp
Performance
8k
tokens
431.9
PP tok/s
24.0
TG tok/s
18967
TTFT (ms)
19.2
Peak mem (GB)
Hardware
Chip
M5 Pro (20c)
Memory
64 GB
GPU Cores
20
Software
oMLX
v0.3.9.dev2
macOS
macOS 26.5
Context
8,192
Performance by Context Length
| Context | PP tok/s | TG tok/s | Peak Mem | |
|---|---|---|---|---|
| 1k | 402.0 | 25.1 | 16.7 GB | view |
| 4k | 462.9 | 24.1 | 18.2 GB | view |
| 8k | 431.9 | 24.0 | 19.2 GB | current |
| 16k | 375.8 | 23.1 | 20.7 GB | view |
| 32k | 351.9 | 19.8 | 23.7 GB | view |
Batching Results
| Batch Size | TG tok/s | Speedup |
|---|---|---|
| 1× | 25.1 | 1.00× |
| 2× | 12.3 | 0.49× |
| 4× | 27.2 | 1.08× |
| 8× | 27.2 | 1.08× |