← Back to community benchmarks
Qwen3.6-27B-MTPLX-Optimized-Speed
Performance
16k
tokens
802.3
PP tok/s
30.0
TG tok/s
20420
TTFT (ms)
20.2
Peak mem (GB)
Hardware
Chip
M5 Max (40c)
Memory
128 GB
GPU Cores
40
Software
oMLX
v0.3.9rc1
macOS
macOS 26.5
Context
16,384
Performance by Context Length
| Context | PP tok/s | TG tok/s | Peak Mem | |
|---|---|---|---|---|
| 1k | 800.4 | 33.1 | 16.3 GB | view |
| 4k | 903.3 | 32.4 | 17.7 GB | view |
| 8k | 842.0 | 31.1 | 18.6 GB | view |
| 16k | 802.3 | 30.0 | 20.2 GB | current |
| 32k | 733.0 | 28.5 | 23.2 GB | view |
| 64k | 615.1 | 25.2 | 29.3 GB | view |
Batching Results
| Batch Size | TG tok/s | Speedup |
|---|---|---|
| 1× | 33.1 | 1.00× |
| 2× | 60.0 | 1.81× |
| 4× | 76.7 | 2.32× |