← Back to community benchmarks
Qwen3.5-9B
Performance
32k
tokens
191.3
PP tok/s
15.4
TG tok/s
171320
TTFT (ms)
8.5
Peak mem (GB)
Hardware
Chip
M4 (10c)
Memory
16 GB
GPU Cores
10
Software
oMLX
v0.2.6
macOS
macOS 26.3
Context
32,768
Performance by Context Length
| Context | PP tok/s | TG tok/s | Peak Mem | |
|---|---|---|---|---|
| 1k | 224.3 | 21.4 | 6.5 GB | view |
| 4k | 225.6 | 20.9 | 6.8 GB | view |
| 8k | 221.8 | 19.7 | 7.0 GB | view |
| 16k | 211.3 | 18.0 | 7.5 GB | view |
| 32k | 191.3 | 15.4 | 8.5 GB | current |
| 64k | 163.0 | 4.9 | 10.6 GB | view |
Batching Results
| Batch Size | TG tok/s | Speedup |
|---|---|---|
| 1× | 21.4 | 1.00× |
| 2× | 40.7 | 1.90× |
| 4× | 51.2 | 2.39× |
| 8× | 52.0 | 2.43× |