← Back to community benchmarks
Qwen3.5-9B-PARO
Performance
8k
tokens
456.4
PP tok/s
46.5
TG tok/s
17948
TTFT (ms)
6.7
Peak mem (GB)
Hardware
Chip
M4 Pro (20c)
Memory
48 GB
GPU Cores
20
Software
oMLX
v0.3.9.dev2
macOS
macOS 26.4.1
Context
8,192
Performance by Context Length
| Context | PP tok/s | TG tok/s | Peak Mem | |
|---|---|---|---|---|
| 1k | 433.7 | 46.6 | 5.7 GB | view |
| 4k | 447.1 | 47.9 | 6.3 GB | view |
| 8k | 456.4 | 46.5 | 6.7 GB | current |
| 16k | 441.8 | 43.1 | 7.7 GB | view |
| 32k | 413.6 | 38.1 | 9.4 GB | view |
| 64k | 363.4 | 30.8 | 13.0 GB | view |
Batching Results
| Batch Size | TG tok/s | Speedup |
|---|---|---|
| 1× | 46.6 | 1.00× |
| 2× | 86.7 | 1.86× |
| 4× | 99.2 | 2.13× |
| 8× | 378.1 | 8.11× |