← Back to community benchmarks
Qwen3.6-27B-oQ8-mtp
Performance
16k
tokens
410.1
PP tok/s
28.4
TG tok/s
39955
TTFT (ms)
32.8
Peak mem (GB)
Hardware
Chip
M3 Ultra (80c)
Memory
512 GB
GPU Cores
80
Software
oMLX
v0.3.9
macOS
macOS 26.3
Context
16,384
Performance by Context Length
| Context | PP tok/s | TG tok/s | Peak Mem | |
|---|---|---|---|---|
| 1k | 388.4 | 31.5 | 28.8 GB | view |
| 4k | 420.7 | 31.1 | 30.3 GB | view |
| 8k | 420.5 | 30.8 | 31.3 GB | view |
| 16k | 410.1 | 28.4 | 32.8 GB | current |
| 32k | 385.3 | 8.8 | 55.7 GB | view |
| 64k | 335.3 | 17.6 | 67.8 GB | view |
Batching Results
| Batch Size | TG tok/s | Speedup |
|---|---|---|
| 1× | 31.5 | 1.00× |
| 2× | 38.6 | 1.23× |
| 4× | 77.2 | 2.45× |