← Back to community benchmarks
Qwen3.5-27B9bit
Performance
32k
tokens
146.3
PP tok/s
12.1
TG tok/s
223963
TTFT (ms)
33.3
Peak mem (GB)
Hardware
Chip
M1 Ultra (64c)
Memory
128 GB
GPU Cores
64
Software
oMLX
v0.2.6
macOS
macOS 26.4
Context
32,768
Performance by Context Length
| Context | PP tok/s | TG tok/s | Peak Mem | |
|---|---|---|---|---|
| 1k | 149.4 | 16.7 | 29.9 GB | view |
| 4k | 156.5 | 16.2 | 30.2 GB | view |
| 8k | 156.3 | 15.3 | 30.6 GB | view |
| 16k | 153.7 | 14.4 | 31.5 GB | view |
| 32k | 146.3 | 12.1 | 33.3 GB | current |
| 64k | 125.8 | 9.9 | 36.9 GB | view |
Batching Results
| Batch Size | TG tok/s | Speedup |
|---|---|---|
| 1× | 16.7 | 1.00× |
| 2× | 31.5 | 1.89× |
| 4× | 39.5 | 2.37× |