← Back to community benchmarks
Qwen3.6-27BoQ8
Performance
16k
tokens
309.4
PP tok/s
21.1
TG tok/s
52948
TTFT (ms)
32.3
Peak mem (GB)
Hardware
Chip
M3 Ultra (60c)
Memory
256 GB
GPU Cores
60
Software
oMLX
v0.3.8
macOS
macOS 26.2
Context
16,384
Performance by Context Length
| Context | PP tok/s | TG tok/s | Peak Mem | |
|---|---|---|---|---|
| 1k | 294.1 | 22.6 | 28.3 GB | view |
| 4k | 318.3 | 22.4 | 29.8 GB | view |
| 8k | 317.6 | 21.9 | 30.8 GB | view |
| 16k | 309.4 | 21.1 | 32.3 GB | current |
| 32k | 291.7 | 19.8 | 35.3 GB | view |
| 64k | 256.1 | 17.6 | 41.4 GB | view |
| 128k | 201.6 | 14.4 | 53.7 GB | view |
| 195k | 165.3 | 12.1 | 68.4 GB | view |
Batching Results
| Batch Size | TG tok/s | Speedup |
|---|---|---|
| 1× | 22.6 | 1.00× |
| 2× | 39.0 | 1.73× |
| 4× | 49.7 | 2.20× |
| 8× | 58.2 | 2.58× |