← Back to community benchmarks
Qwen3.5-27B
Performance
16k
tokens
85.0
PP tok/s
12.2
TG tok/s
192850
TTFT (ms)
18.3
Peak mem (GB)
Hardware
Chip
M1 Max (32c)
Memory
32 GB
GPU Cores
32
Software
oMLX
v0.2.6
macOS
macOS 26.2
Context
16,384
Performance by Context Length
| Context | PP tok/s | TG tok/s | Peak Mem | |
|---|---|---|---|---|
| 1k | 89.8 | 19.1 | 16.6 GB | view |
| 4k | 90.5 | 17.8 | 17.0 GB | view |
| 8k | 89.6 | 16.3 | 17.4 GB | view |
| 16k | 85.0 | 12.2 | 18.3 GB | current |
| 32k | 74.8 | 9.5 | 20.1 GB | view |
Batching Results
| Batch Size | TG tok/s | Speedup |
|---|---|---|
| 1× | 19.1 | 1.00× |
| 2× | 17.6 | 0.92× |
| 4× | 18.7 | 0.98× |
| 8× | 16.0 | 0.84× |