← Back to community benchmarks
Qwen3.5-4B
Performance
4k
tokens
406.3
PP tok/s
36.7
TG tok/s
10081
TTFT (ms)
4.2
Peak mem (GB)
Hardware
Chip
M4 (10c)
Memory
16 GB
GPU Cores
10
Software
oMLX
v0.2.9
macOS
macOS 26.2
Context
4,096
Performance by Context Length
| Context | PP tok/s | TG tok/s | Peak Mem | |
|---|---|---|---|---|
| 1k | 408.9 | 38.8 | 3.9 GB | view |
| 4k | 406.3 | 36.7 | 4.2 GB | current |
| 8k | 394.1 | 34.1 | 4.6 GB | view |
| 16k | 369.2 | 25.1 | 4.8 GB | view |
| 32k | 327.1 | 16.6 | 5.9 GB | view |
| 64k | 254.0 | 10.4 | 8.0 GB | view |
Batching Results
| Batch Size | TG tok/s | Speedup |
|---|---|---|
| 1× | 38.8 | 1.00× |
| 2× | 72.9 | 1.88× |
| 4× | 89.3 | 2.30× |
| 8× | 89.9 | 2.32× |