← Back to community benchmarks
Qwen3.5-122B-A10B
Performance
64k
tokens
359.8
PP tok/s
21.9
TG tok/s
182157
TTFT (ms)
76.1
Peak mem (GB)
Hardware
Chip
M4 Max (40c)
Memory
128 GB
GPU Cores
40
Software
oMLX
v0.3.2
macOS
macOS 15.7.5
Context
65,536
Performance by Context Length
| Context | PP tok/s | TG tok/s | Peak Mem | |
|---|---|---|---|---|
| 1k | 455.5 | 51.7 | 65.6 GB | view |
| 4k | 521.7 | 49.2 | 67.1 GB | view |
| 8k | 518.6 | 46.9 | 67.7 GB | view |
| 16k | 499.1 | 39.7 | 68.9 GB | view |
| 32k | 441.0 | 31.6 | 71.3 GB | view |
| 64k | 359.8 | 21.9 | 76.1 GB | current |
| 128k | 247.7 | 13.6 | 85.8 GB | view |
| 195k | 163.1 | 8.8 | 95.9 GB | view |
Batching Results
| Batch Size | TG tok/s | Speedup |
|---|---|---|
| 1× | 51.7 | 1.00× |
| 2× | 83.2 | 1.61× |