← Back to community benchmarks
Qwen3.6-27B-MXFP4-MTP
Performance
16k
tokens
159.3
PP tok/s
7.5
TG tok/s
102833
TTFT (ms)
18.9
Peak mem (GB)
Hardware
Chip
M5 (10c)
Memory
32 GB
GPU Cores
10
Software
oMLX
v0.3.9rc1
macOS
macOS 26.4
Context
16,384
Performance by Context Length
| Context | PP tok/s | TG tok/s | Peak Mem | |
|---|---|---|---|---|
| 1k | 196.5 | 8.8 | 15.0 GB | view |
| 4k | 202.2 | 8.7 | 16.4 GB | view |
| 8k | 189.7 | 8.5 | 17.4 GB | view |
| 16k | 159.3 | 7.5 | 18.9 GB | current |
Batching Results
| Batch Size | TG tok/s | Speedup |
|---|---|---|
| 1× | 8.8 | 1.00× |
| 2× | 15.1 | 1.72× |
| 4× | 25.8 | 2.93× |