← Back to community benchmarks
Qwen3.5-27B
Performance
16k
tokens
176.2
PP tok/s
25.3
TG tok/s
92980
TTFT (ms)
18.3
Peak mem (GB)
Hardware
Chip
M2 Ultra (60c)
Memory
64 GB
GPU Cores
60
Software
oMLX
v0.2.6
macOS
macOS 26.3.1
Context
16,384
Performance by Context Length
| Context | PP tok/s | TG tok/s | Peak Mem | |
|---|---|---|---|---|
| 1k | 180.4 | 32.2 | 16.6 GB | view |
| 4k | 182.4 | 30.4 | 17.0 GB | view |
| 8k | 180.9 | 28.4 | 17.4 GB | view |
| 16k | 176.2 | 25.3 | 18.3 GB | current |
Batching Results
| Batch Size | TG tok/s | Speedup |
|---|---|---|
| 1× | 32.2 | 1.00× |
| 2× | 40.4 | 1.25× |
| 4× | 45.1 | 1.40× |