← Back to community benchmarks
Qwen3.5-9B
Performance
16k
tokens
211.9
PP tok/s
17.6
TG tok/s
77305
TTFT (ms)
7.5
Peak mem (GB)
Hardware
Chip
M4 (10c)
Memory
16 GB
GPU Cores
10
Software
oMLX
v0.2.6
macOS
macOS 26.3
Context
16,384
Performance by Context Length
| Context | PP tok/s | TG tok/s | Peak Mem | |
|---|---|---|---|---|
| 1k | 223.6 | 21.3 | 6.5 GB | view |
| 4k | 225.5 | 20.8 | 6.8 GB | view |
| 8k | 220.2 | 19.8 | 7.0 GB | view |
| 16k | 211.9 | 17.6 | 7.5 GB | current |
| 32k | 187.3 | 12.2 | 8.5 GB | view |
| 64k | 160.7 | 10.8 | 10.6 GB | view |
Batching Results
| Batch Size | TG tok/s | Speedup |
|---|---|---|
| 1× | 21.3 | 1.00× |
| 2× | 41.6 | 1.95× |
| 4× | 48.6 | 2.28× |
| 8× | 51.1 | 2.40× |