← Back to community benchmarks
Qwen3.6-27B
Performance
1k
tokens
900.7
PP tok/s
31.5
TG tok/s
1137
TTFT (ms)
15.8
Peak mem (GB)
Hardware
Chip
M5 Max (40c)
Memory
128 GB
GPU Cores
40
Software
oMLX
v0.5.1
macOS
macOS 26.5
Context
1,024
Performance by Context Length
| Context | PP tok/s | TG tok/s | Peak Mem | |
|---|---|---|---|---|
| 1k | 900.7 | 31.5 | 15.8 GB | current |
| 4k | 896.7 | 31.1 | 17.3 GB | view |
| 8k | 835.2 | 29.8 | 18.6 GB | view |
| 16k | 693.3 | 28.2 | 20.4 GB | view |
| 32k | 601.3 | 23.9 | 23.6 GB | view |
| 64k | 419.6 | 19.3 | 29.1 GB | view |
Batching Results
| Batch Size | TG tok/s | Speedup |
|---|---|---|
| 1× | 31.5 | 1.00× |
| 2× | 41.1 | 1.30× |
| 4× | 63.6 | 2.02× |
| 8× | 69.8 | 2.22× |