← Back to community benchmarks
Qwen3.5-122B-A10B-oQ4
Performance
16k
tokens
1,335
PP tok/s
47.7
TG tok/s
12271
TTFT (ms)
70.4
Peak mem (GB)
Hardware
Chip
M5 Max (40c)
Memory
128 GB
GPU Cores
40
Software
oMLX
v0.3.0
macOS
macOS 26.4
Context
16,384
Performance by Context Length
| Context | PP tok/s | TG tok/s | Peak Mem | |
|---|---|---|---|---|
| 1k | 1,032 | 57.3 | 67.1 GB | view |
| 4k | 1,470 | 55.2 | 68.6 GB | view |
| 8k | 1,436 | 52.5 | 69.2 GB | view |
| 16k | 1,335 | 47.7 | 70.4 GB | current |
| 32k | 1,112 | 39.6 | 72.8 GB | view |
| 64k | 868.8 | 30.6 | 77.6 GB | view |
Batching Results
| Batch Size | TG tok/s | Speedup |
|---|---|---|
| 1× | 57.3 | 1.00× |
| 2× | 87.0 | 1.52× |
| 4× | 111.1 | 1.94× |
| 8× | 138.9 | 2.42× |