← Back to community benchmarks
Qwen3.6-27B-oQ4-fp16-mtp
Performance
16k
tokens
655.3
PP tok/s
38.8
TG tok/s
25004
TTFT (ms)
20.7
Peak mem (GB)
Hardware
Chip
M5 Max (40c)
Memory
128 GB
GPU Cores
40
Software
oMLX
v0.3.9
macOS
macOS 26.4
Context
16,384
Performance by Context Length
| Context | PP tok/s | TG tok/s | Peak Mem | |
|---|---|---|---|---|
| 1k | 762.1 | 45.2 | 16.7 GB | view |
| 4k | 818.1 | 43.4 | 18.2 GB | view |
| 8k | 734.1 | 41.8 | 19.2 GB | view |
| 16k | 655.3 | 38.8 | 20.7 GB | current |
| 32k | 481.2 | 20.0 | 23.7 GB | view |
Batching Results
| Batch Size | TG tok/s | Speedup |
|---|---|---|
| 1× | 45.2 | 1.00× |
| 2× | 28.1 | 0.62× |
| 4× | 36.8 | 0.81× |
| 8× | 43.8 | 0.97× |