← Back to community benchmarks
Qwen3.5-122B-A10B
Performance
32k
tokens
743.3
PP tok/s
29.8
TG tok/s
44083
TTFT (ms)
125.9
Peak mem (GB)
Hardware
Chip
M3 Ultra (80c)
Memory
512 GB
GPU Cores
80
Software
oMLX
v0.2.6
macOS
macOS 26.3
Context
32,768
Performance by Context Length
| Context | PP tok/s | TG tok/s | Peak Mem | |
|---|---|---|---|---|
| 1k | 858.8 | 40.4 | 123.2 GB | view |
| 4k | 906.1 | 39.1 | 123.5 GB | view |
| 8k | 895.2 | 37.6 | 123.9 GB | view |
| 16k | 843.1 | 35.2 | 124.5 GB | view |
| 32k | 743.3 | 29.8 | 125.9 GB | current |
Batching Results
| Batch Size | TG tok/s | Speedup |
|---|---|---|
| 1× | 40.4 | 1.00× |
| 2× | 69.2 | 1.71× |
| 4× | 95.0 | 2.35× |