← Back to community benchmarks
Qwen3.5-9B
Performance
64k
tokens
668.6
PP tok/s
33.9
TG tok/s
98023
TTFT (ms)
14.8
Peak mem (GB)
Hardware
Chip
M2 Ultra (76c)
Memory
128 GB
GPU Cores
76
Software
oMLX
v0.2.6
macOS
macOS 26.3
Context
65,536
Performance by Context Length
| Context | PP tok/s | TG tok/s | Peak Mem | |
|---|---|---|---|---|
| 1k | 794.0 | 64.3 | 10.7 GB | view |
| 4k | 804.0 | 62.6 | 11.0 GB | view |
| 8k | 797.4 | 59.7 | 11.2 GB | view |
| 16k | 778.7 | 54.0 | 11.7 GB | view |
| 32k | 742.0 | 46.4 | 12.7 GB | view |
| 64k | 668.6 | 33.9 | 14.8 GB | current |
Batching Results
| Batch Size | TG tok/s | Speedup |
|---|---|---|
| 1× | 64.3 | 1.00× |
| 2× | 109.3 | 1.70× |
| 4× | 150.8 | 2.35× |
| 8× | 190.1 | 2.96× |