← Back to community benchmarks
Qwen3.5-VL-122B-A10Bcrk
Performance
16k
tokens
644.4
PP tok/s
38.7
TG tok/s
25425
TTFT (ms)
67.7
Peak mem (GB)
Hardware
Chip
M3 Ultra (60c)
Memory
96 GB
GPU Cores
60
Software
oMLX
v0.2.9
macOS
macOS 26.3.1
Context
16,384
Performance by Context Length
| Context | PP tok/s | TG tok/s | Peak Mem | |
|---|---|---|---|---|
| 1k | 624.3 | 50.6 | 66.4 GB | view |
| 4k | 691.7 | 48.3 | 66.7 GB | view |
| 8k | 688.1 | 44.9 | 67.0 GB | view |
| 16k | 644.4 | 38.7 | 67.7 GB | current |
| 32k | 568.2 | 29.8 | 69.1 GB | view |
| 64k | 450.3 | 21.3 | 71.9 GB | view |
Batching Results
| Batch Size | TG tok/s | Speedup |
|---|---|---|
| 1× | 50.6 | 1.00× |
| 2× | 50.2 | 0.99× |
| 4× | 120.4 | 2.38× |
| 8× | 124.9 | 2.47× |