← Back to community benchmarks
Qwen3.6-27B-mxfp8
Performance
128k
tokens
345.3
PP tok/s
12.1
TG tok/s
379544
TTFT (ms)
48.8
Peak mem (GB)
Hardware
Chip
M5 Max (40c)
Memory
128 GB
GPU Cores
40
Software
oMLX
v0.3.8
macOS
macOS 26.4.1
Context
131,072
Performance by Context Length
| Context | PP tok/s | TG tok/s | Peak Mem | |
|---|---|---|---|---|
| 1k | 703.7 | 19.3 | 27.6 GB | view |
| 4k | 753.2 | 19.1 | 29.0 GB | view |
| 8k | 677.5 | 18.7 | 29.6 GB | view |
| 16k | 628.5 | 17.5 | 30.9 GB | view |
| 32k | 604.0 | 17.1 | 33.4 GB | view |
| 64k | 510.6 | 15.2 | 38.4 GB | view |
| 128k | 345.3 | 12.1 | 48.8 GB | current |
| 195k | 270.2 | 10.4 | 59.5 GB | view |
Batching Results
| Batch Size | TG tok/s | Speedup |
|---|---|---|
| 1× | 19.3 | 1.00× |
| 2× | 25.9 | 1.34× |
| 4× | 32.3 | 1.67× |
| 8× | 35.8 | 1.85× |