← Back to community benchmarks
Qwen3.6-27B-oQ4-fp16-mtp
Performance
1k
tokens
275.1
PP tok/s
36.1
TG tok/s
3722
TTFT (ms)
16.7
Peak mem (GB)
Hardware
Chip
M2 Ultra (60c)
Memory
64 GB
GPU Cores
60
Software
oMLX
v0.3.9rc1
macOS
macOS 26.3.1
Context
1,024
Performance by Context Length
| Context | PP tok/s | TG tok/s | Peak Mem | |
|---|---|---|---|---|
| 1k | 275.1 | 36.1 | 16.7 GB | current |
| 4k | 295.2 | 36.5 | 18.2 GB | view |