← Back to community benchmarks
qwen3.5-122b-a10bmxfp8
Performance
16k
tokens
591.7
PP tok/s
25.8
TG tok/s
27688
TTFT (ms)
123.8
Peak mem (GB)
Hardware
Chip
M2 Ultra (76c)
Memory
192 GB
GPU Cores
76
Software
oMLX
v0.2.6
macOS
macOS 26.3
Context
16,384
Performance by Context Length
| Context | PP tok/s | TG tok/s | Peak Mem | |
|---|---|---|---|---|
| 1k | 570.7 | 37.5 | 121.2 GB | view |
| 8k | 633.7 | 32.0 | 122.5 GB | view |
| 16k | 591.7 | 25.8 | 123.8 GB | current |
Batching Results
| Batch Size | TG tok/s | Speedup |
|---|---|---|
| 1× | 37.5 | 1.00× |
| 2× | 58.7 | 1.57× |
| 4× | 76.2 | 2.03× |