← Back to community benchmarks
Qwen3.5-27B
Performance
4k
tokens
228.9
PP tok/s
31.3
TG tok/s
17897
TTFT (ms)
17.0
Peak mem (GB)
Hardware
Chip
M2 Ultra (76c)
Memory
128 GB
GPU Cores
76
Software
oMLX
v0.2.6
macOS
macOS 26.3
Context
4,096
Performance by Context Length
| Context | PP tok/s | TG tok/s | Peak Mem | |
|---|---|---|---|---|
| 1k | 226.5 | 33.4 | 16.6 GB | view |
| 4k | 228.9 | 31.3 | 17.0 GB | current |
| 8k | 227.0 | 29.3 | 17.4 GB | view |
| 16k | 222.3 | 26.6 | 18.3 GB | view |
| 32k | 212.9 | 21.9 | 20.1 GB | view |
| 64k | 192.2 | 16.8 | 23.6 GB | view |
Batching Results
| Batch Size | TG tok/s | Speedup |
|---|---|---|
| 1× | 33.4 | 1.00× |
| 2× | 47.7 | 1.43× |
| 4× | 53.5 | 1.60× |
| 8× | 58.9 | 1.76× |