← Back to community benchmarks
Qwen3.5-9B
Performance
64k
tokens
166.1
PP tok/s
6.1
TG tok/s
394640
TTFT (ms)
10.6
Peak mem (GB)
Hardware
Chip
M4 (10c)
Memory
16 GB
GPU Cores
10
Software
oMLX
v0.2.6
macOS
macOS 26.3.1
Context
65,536
Performance by Context Length
| Context | PP tok/s | TG tok/s | Peak Mem | |
|---|---|---|---|---|
| 1k | 222.7 | 19.8 | 6.5 GB | view |
| 4k | 224.2 | 19.1 | 6.8 GB | view |
| 8k | 221.2 | 16.5 | 7.0 GB | view |
| 16k | 210.3 | 12.9 | 7.5 GB | view |
| 32k | 193.9 | 9.7 | 8.5 GB | view |
| 64k | 166.1 | 6.1 | 10.6 GB | current |
Batching Results
| Batch Size | TG tok/s | Speedup |
|---|---|---|
| 1× | 19.8 | 1.00× |
| 2× | 36.2 | 1.83× |
| 4× | 51.3 | 2.59× |
| 8× | 48.9 | 2.47× |