← Back to community benchmarks
Qwen3.5-122B-A10B
Performance
16k
tokens
2,033
PP tok/s
35.4
TG tok/s
8059
TTFT (ms)
125.4
Peak mem (GB)
Hardware
Chip
M3 Ultra (60c)
Memory
256 GB
GPU Cores
60
Software
oMLX
v0.2.24
macOS
macOS 26.3
Context
16,384
Performance by Context Length
| Context | PP tok/s | TG tok/s | Peak Mem | |
|---|---|---|---|---|
| 1k | 676.4 | 36.0 | 124.3 GB | view |
| 4k | 743.2 | 34.8 | 125.8 GB | view |
| 8k | 729.4 | 33.5 | 126.4 GB | view |
| 16k | 2,033 | 35.4 | 125.4 GB | current |
| 32k | 1,938 | 34.5 | 126.1 GB | view |
| 64k | 1,743 | 33.1 | 126.9 GB | view |
| 128k | 1,403 | 31.2 | 129.7 GB | view |
Batching Results
| Batch Size | TG tok/s | Speedup |
|---|---|---|
| 1× | 36.0 | 1.00× |
| 2× | 56.4 | 1.57× |
| 4× | 78.0 | 2.17× |
| 8× | 115.0 | 3.19× |