← Back to community benchmarks
Qwen3.6-27B-oQ4-mtp
Performance
8k
tokens
393.8
PP tok/s
39.5
TG tok/s
20803
TTFT (ms)
26.7
Peak mem (GB)
Hardware
Chip
M3 Ultra (80c)
Memory
512 GB
GPU Cores
80
Software
oMLX
v0.3.9.dev2
macOS
macOS 26.3.1
Context
8,192
Performance by Context Length
| Context | PP tok/s | TG tok/s | Peak Mem | |
|---|---|---|---|---|
| 1k | 363.0 | 41.6 | 24.2 GB | view |
| 4k | 392.5 | 41.9 | 25.6 GB | view |
| 8k | 393.8 | 39.5 | 26.7 GB | current |
Batching Results
| Batch Size | TG tok/s | Speedup |
|---|---|---|
| 1× | 41.6 | 1.00× |
| 2× | 58.3 | 1.40× |
| 4× | 78.3 | 1.88× |
| 8× | 88.1 | 2.12× |