← Back to community benchmarks
Qwen3.5-4B-Uncensored
Performance
32k
tokens
491.7
PP tok/s
20.4
TG tok/s
66644
TTFT (ms)
6.7
Peak mem (GB)
Hardware
Chip
M5 (10c)
Memory
24 GB
GPU Cores
10
Software
oMLX
v0.3.2
macOS
macOS 26.4
Context
32,768
Performance by Context Length
| Context | PP tok/s | TG tok/s | Peak Mem | |
|---|---|---|---|---|
| 1k | 929.1 | 46.6 | 3.3 GB | view |
| 4k | 959.2 | 43.8 | 4.0 GB | view |
| 8k | 769.7 | 33.2 | 4.4 GB | view |
| 16k | 605.1 | 23.2 | 5.1 GB | view |
| 32k | 491.7 | 20.4 | 6.7 GB | current |
| 64k | 385.2 | 13.9 | 9.8 GB | view |
| 128k | 235.6 | 8.0 | 16.1 GB | view |
Batching Results
| Batch Size | TG tok/s | Speedup |
|---|---|---|
| 1× | 46.6 | 1.00× |
| 2× | 45.1 | 0.97× |