← Back to community benchmarks
Qwen3.6-27B-oQ4-mtp
Performance
16k
tokens
241.3
PP tok/s
35.0
TG tok/s
67890
TTFT (ms)
20.7
Peak mem (GB)
Hardware
Chip
M4 Max (40c)
Memory
128 GB
GPU Cores
40
Software
oMLX
v0.3.9.dev2
macOS
macOS 26.4.1
Context
16,384
Performance by Context Length
| Context | PP tok/s | TG tok/s | Peak Mem | |
|---|---|---|---|---|
| 1k | 241.9 | 39.1 | 16.7 GB | view |
| 4k | 249.2 | 39.9 | 18.2 GB | view |
| 8k | 247.9 | 38.1 | 19.2 GB | view |
| 16k | 241.3 | 35.0 | 20.7 GB | current |
| 32k | 227.8 | 32.6 | 23.7 GB | view |
Batching Results
| Batch Size | TG tok/s | Speedup |
|---|---|---|
| 1× | 39.1 | 1.00× |
| 2× | 49.8 | 1.27× |
| 4× | 55.5 | 1.42× |
| 8× | 112.1 | 2.87× |