← Back to community benchmarks
Qwen3.5-9B-oQ4-mtp
Performance
32k
tokens
634.9
PP tok/s
54.9
TG tok/s
51613
TTFT (ms)
9.8
Peak mem (GB)
Hardware
Chip
M3 Max (40c)
Memory
64 GB
GPU Cores
40
Software
oMLX
v0.3.9.dev2
macOS
macOS 26.5
Context
32,768
Performance by Context Length
| Context | PP tok/s | TG tok/s | Peak Mem | |
|---|---|---|---|---|
| 1k | 671.2 | 68.3 | 5.9 GB | view |
| 4k | 746.4 | 67.3 | 6.6 GB | view |
| 8k | 745.7 | 65.2 | 7.0 GB | view |
| 16k | 675.0 | 61.8 | 8.0 GB | view |
| 32k | 634.9 | 54.9 | 9.8 GB | current |
Batching Results
| Batch Size | TG tok/s | Speedup |
|---|---|---|
| 1× | 68.3 | 1.00× |
| 2× | 118.7 | 1.74× |
| 4× | 142.9 | 2.09× |