← Back to community benchmarks
Qwen3.6-27B-oQ8-mtp
Performance
8k
tokens
318.3
PP tok/s
30.7
TG tok/s
25736
TTFT (ms)
31.3
Peak mem (GB)
Hardware
Chip
M3 Ultra (60c)
Memory
256 GB
GPU Cores
60
Software
oMLX
v0.3.9.dev2
macOS
macOS 26.4.1
Context
8,192
Performance by Context Length
| Context | PP tok/s | TG tok/s | Peak Mem | |
|---|---|---|---|---|
| 1k | 294.9 | 32.6 | 28.8 GB | view |
| 4k | 319.5 | 32.0 | 30.3 GB | view |
| 8k | 318.3 | 30.7 | 31.3 GB | current |
| 16k | 310.0 | 28.6 | 32.8 GB | view |
| 32k | 292.0 | 27.0 | 35.8 GB | view |
| 64k | 256.7 | 23.4 | 41.8 GB | view |
| 128k | 203.1 | 18.4 | 54.2 GB | view |
Batching Results
| Batch Size | TG tok/s | Speedup |
|---|---|---|
| 1× | 32.6 | 1.00× |
| 2× | 39.3 | 1.21× |
| 4× | 50.0 | 1.53× |