← Back to community benchmarks
Qwen3.6-27B
Performance
32k
tokens
81.4
PP tok/s
8.1
TG tok/s
402410
TTFT (ms)
20.9
Peak mem (GB)
Hardware
Chip
M3 Max (40c)
Memory
128 GB
GPU Cores
40
Software
oMLX
v0.3.6
macOS
macOS 26.4.1
Context
32,768
Performance by Context Length
| Context | PP tok/s | TG tok/s | Peak Mem | |
|---|---|---|---|---|
| 1k | 99.8 | 9.1 | 15.1 GB | view |
| 4k | 99.0 | 12.5 | 16.5 GB | view |
| 8k | 129.1 | 12.9 | 17.1 GB | view |
| 16k | 68.2 | 5.4 | 18.4 GB | view |
| 32k | 81.4 | 8.1 | 20.9 GB | current |
Batching Results
| Batch Size | TG tok/s | Speedup |
|---|---|---|
| 1× | 9.1 | 1.00× |
| 2× | 13.7 | 1.51× |
| 4× | 16.8 | 1.85× |