← Back to community benchmarks
Qwen3.5-4B
Performance
32k
tokens
299.7
PP tok/s
20.3
TG tok/s
109323
TTFT (ms)
5.9
Peak mem (GB)
Hardware
Chip
M4 (10c)
Memory
16 GB
GPU Cores
10
Software
oMLX
v0.2.6
macOS
macOS 26.3
Context
32,768
Performance by Context Length
| Context | PP tok/s | TG tok/s | Peak Mem | |
|---|---|---|---|---|
| 1k | 374.9 | 33.7 | 3.9 GB | view |
| 4k | 373.3 | 32.0 | 4.2 GB | view |
| 8k | 364.6 | 29.5 | 4.6 GB | view |
| 16k | 340.6 | 25.2 | 4.8 GB | view |
| 32k | 299.7 | 20.3 | 5.9 GB | current |
| 64k | 231.3 | 13.6 | 8.0 GB | view |
Batching Results
| Batch Size | TG tok/s | Speedup |
|---|---|---|
| 1× | 33.7 | 1.00× |
| 2× | 63.8 | 1.89× |
| 4× | 82.6 | 2.45× |
| 8× | 85.0 | 2.52× |