← Back to community benchmarks
Qwen3.5-VL-122B-A10B-4bitCRACK
Performance
32k
tokens
309.2
PP tok/s
23.2
TG tok/s
105991
TTFT (ms)
71.2
Peak mem (GB)
Hardware
Chip
M3 Max (40c)
Memory
128 GB
GPU Cores
40
Software
oMLX
v0.3.0
macOS
macOS 26.4
Context
32,768
Performance by Context Length
| Context | PP tok/s | TG tok/s | Peak Mem | |
|---|---|---|---|---|
| 1k | 414.9 | 43.7 | 65.6 GB | view |
| 4k | 470.4 | 40.4 | 67.0 GB | view |
| 8k | 411.3 | 36.3 | 67.7 GB | view |
| 16k | 360.8 | 33.1 | 68.9 GB | view |
| 32k | 309.2 | 23.2 | 71.2 GB | current |
| 64k | 140.1 | 12.1 | 76.1 GB | view |
| 128k | 118.9 | 7.6 | 85.7 GB | view |
Batching Results
| Batch Size | TG tok/s | Speedup |
|---|---|---|
| 1× | 43.7 | 1.00× |
| 2× | 34.6 | 0.79× |
| 4× | 41.7 | 0.95× |