← Back to community benchmarks
Qwen3.6-35B-A3B-uncensored-heretic-Native-MTP-Preserved
Performance
32k
tokens
1,121
PP tok/s
70.1
TG tok/s
29234
TTFT (ms)
37.2
Peak mem (GB)
Hardware
Chip
M4 Max (40c)
Memory
128 GB
GPU Cores
40
Software
oMLX
v0.5.1
macOS
macOS 26.5.2
Context
32,768
Performance by Context Length
| Context | PP tok/s | TG tok/s | Peak Mem | |
|---|---|---|---|---|
| 1k | 1,734 | 83.4 | 35.4 GB | view |
| 4k | 1,837 | 81.6 | 36.1 GB | view |
| 8k | 1,648 | 78.1 | 36.3 GB | view |
| 16k | 1,392 | 75.4 | 36.5 GB | view |
| 32k | 1,121 | 70.1 | 37.2 GB | current |
| 64k | 849.2 | 60.1 | 37.7 GB | view |
| 128k | 535.4 | 39.5 | 38.9 GB | view |
| 195k | 347.7 | 31.5 | 40.5 GB | view |
Batching Results
| Batch Size | TG tok/s | Speedup |
|---|---|---|
| 1× | 83.4 | 1.00× |
| 2× | 135.6 | 1.63× |
| 4× | 172.4 | 2.07× |