← Back to community benchmarks
Qwen3.5-122B-A10B
Performance
32k
tokens
744.4
PP tok/s
31.4
TG tok/s
44021
TTFT (ms)
153.8
Peak mem (GB)
Hardware
Chip
M3 Ultra (80c)
Memory
512 GB
GPU Cores
80
Software
oMLX
v0.3.4
macOS
macOS 15.4.1
Context
32,768
Performance by Context Length
| Context | PP tok/s | TG tok/s | Peak Mem | |
|---|---|---|---|---|
| 1k | 735.1 | 41.9 | 148.3 GB | view |
| 4k | 911.1 | 40.6 | 149.6 GB | view |
| 8k | 912.5 | 39.1 | 150.2 GB | view |
| 16k | 862.1 | 36.3 | 151.4 GB | view |
| 32k | 744.4 | 31.4 | 153.8 GB | current |
| 64k | 548.7 | 25.1 | 158.7 GB | view |
Batching Results
| Batch Size | TG tok/s | Speedup |
|---|---|---|
| 1× | 41.9 | 1.00× |
| 2× | 65.9 | 1.57× |
| 4× | 89.0 | 2.12× |
| 8× | 121.0 | 2.89× |