← Back to community benchmarks
Qwen3.5-2B-Polaris-HighIQ-INSTRUCT
Performance
16k
tokens
1,703
PP tok/s
57.4
TG tok/s
9619
TTFT (ms)
5.2
Peak mem (GB)
Hardware
Chip
M4 Pro (16c)
Memory
24 GB
GPU Cores
16
Software
oMLX
v0.3.6
macOS
macOS 26.4.1
Context
16,384
Performance by Context Length
| Context | PP tok/s | TG tok/s | Peak Mem | |
|---|---|---|---|---|
| 1k | 1,373 | 62.1 | 4.2 GB | view |
| 4k | 1,705 | 59.2 | 4.7 GB | view |
| 8k | 1,737 | 59.5 | 4.9 GB | view |
| 16k | 1,703 | 57.4 | 5.2 GB | current |
| 32k | 1,530 | 51.6 | 5.9 GB | view |
| 64k | 1,275 | 43.2 | 7.3 GB | view |
Batching Results
| Batch Size | TG tok/s | Speedup |
|---|---|---|
| 1× | 62.1 | 1.00× |
| 2× | 51.5 | 0.83× |
| 4× | 100.3 | 1.62× |
| 8× | 188.3 | 3.03× |