← Back to community benchmarks
Qwen3.5-4B
Performance
8k
tokens
390.4
PP tok/s
35.5
TG tok/s
20986
TTFT (ms)
4.4
Peak mem (GB)
Hardware
Chip
M4 (10c)
Memory
16 GB
GPU Cores
10
Software
oMLX
v0.2.6
macOS
macOS 26.3
Context
8,192
Performance by Context Length
| Context | PP tok/s | TG tok/s | Peak Mem | |
|---|---|---|---|---|
| 1k | 402.8 | 39.8 | 3.8 GB | view |
| 4k | 400.1 | 38.3 | 4.0 GB | view |
| 8k | 390.4 | 35.5 | 4.4 GB | current |
Batching Results
| Batch Size | TG tok/s | Speedup |
|---|---|---|
| 1× | 39.8 | 1.00× |
| 2× | 76.2 | 1.91× |
| 4× | 102.9 | 2.59× |
| 8× | 105.1 | 2.64× |