← Back to community benchmarks
Qwen3.5-9B
Performance
16k
tokens
774.4
PP tok/s
76.6
TG tok/s
21158
TTFT (ms)
7.6
Peak mem (GB)
Hardware
Chip
M2 Ultra (76c)
Memory
192 GB
GPU Cores
76
Software
oMLX
v0.2.6
macOS
macOS 26.3.1
Context
16,384
Performance by Context Length
| Context | PP tok/s | TG tok/s | Peak Mem | |
|---|---|---|---|---|
| 1k | 727.6 | 96.7 | 6.7 GB | view |
| 4k | 789.6 | 93.2 | 7.0 GB | view |
| 8k | 792.6 | 87.6 | 7.1 GB | view |
| 16k | 774.4 | 76.6 | 7.6 GB | current |
| 32k | 738.1 | 61.9 | 8.7 GB | view |
| 64k | 665.7 | 44.3 | 10.8 GB | view |
Batching Results
| Batch Size | TG tok/s | Speedup |
|---|---|---|
| 1× | 96.7 | 1.00× |
| 2× | 135.8 | 1.40× |
| 4× | 161.6 | 1.67× |
| 8× | 182.6 | 1.89× |