← Back to community benchmarks
Qwen3.5-122B-A10B-5bit
Performance
16k
tokens
379.5
PP tok/s
25.4
TG tok/s
43168
TTFT (ms)
83.1
Peak mem (GB)
Hardware
Chip
M3 Max (40c)
Memory
128 GB
GPU Cores
40
Software
oMLX
v0.3.0rc1
macOS
macOS 26.3.1
Context
16,384
Performance by Context Length
| Context | PP tok/s | TG tok/s | Peak Mem | |
|---|---|---|---|---|
| 1k | 421.8 | 40.4 | 79.8 GB | view |
| 4k | 479.2 | 38.2 | 81.3 GB | view |
| 8k | 434.1 | 35.5 | 81.9 GB | view |
| 16k | 379.5 | 25.4 | 83.1 GB | current |
| 32k | 354.4 | 23.7 | 85.5 GB | view |
| 64k | 289.8 | 17.3 | 90.3 GB | view |
| 128k | 199.9 | 7.2 | 100.0 GB | view |
Batching Results
| Batch Size | TG tok/s | Speedup |
|---|---|---|
| 1× | 40.4 | 1.00× |
| 2× | 52.1 | 1.29× |
| 4× | 60.0 | 1.49× |