← Back to community benchmarks
Qwen3.6-35B-A3B-uncensored-heretic-Native-MTP-Preserved
Performance
16k
tokens
3,533
PP tok/s
64.1
TG tok/s
4638
TTFT (ms)
67.6
Peak mem (GB)
Hardware
Chip
M5 Max (40c)
Memory
128 GB
GPU Cores
40
Software
oMLX
v0.5.3
macOS
macOS 27.0
Context
16,384
Performance by Context Length
| Context | PP tok/s | TG tok/s | Peak Mem | |
|---|---|---|---|---|
| 1k | 3,060 | 71.0 | 65.8 GB | view |
| 4k | 4,276 | 70.4 | 66.6 GB | view |
| 8k | 3,882 | 69.2 | 66.9 GB | view |
| 16k | 3,533 | 64.1 | 67.6 GB | current |
| 32k | 2,883 | 63.0 | 68.9 GB | view |
| 64k | 2,188 | 57.0 | 71.5 GB | view |
| 128k | 1,348 | 47.9 | 76.9 GB | view |
Batching Results
| Batch Size | TG tok/s | Speedup |
|---|---|---|
| 1× | 71.0 | 1.00× |
| 2× | 67.8 | 0.95× |
| 4× | 119.4 | 1.68× |