← Back to community benchmarks
Qwen3.5-27B
Performance
8k
tokens
319.7
PP tok/s
33.6
TG tok/s
25625
TTFT (ms)
17.4
Peak mem (GB)
Hardware
Chip
M3 Ultra (60c)
Memory
256 GB
GPU Cores
60
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 | 322.7 | 38.0 | 16.6 GB | view |
| 4k | 324.2 | 35.3 | 17.0 GB | view |
| 8k | 319.7 | 33.6 | 17.4 GB | current |
| 16k | 310.0 | 30.0 | 18.3 GB | view |
| 32k | 291.3 | 24.9 | 20.1 GB | view |
| 64k | 258.7 | 18.1 | 23.6 GB | view |
Batching Results
| Batch Size | TG tok/s | Speedup |
|---|---|---|
| 1× | 38.0 | 1.00× |
| 2× | 62.0 | 1.63× |
| 4× | 71.5 | 1.88× |
| 8× | 80.5 | 2.12× |