← Back to community benchmarks
Qwen3.6-27B-MXFP8-MTP
Performance
16k
tokens
58.0
PP tok/s
3.6
TG tok/s
282325
TTFT (ms)
37.4
Peak mem (GB)
Hardware
Chip
M3 Max (40c)
Memory
128 GB
GPU Cores
40
Software
oMLX
v0.3.9
macOS
macOS 26.4.1
Context
16,384
Performance by Context Length
| Context | PP tok/s | TG tok/s | Peak Mem | |
|---|---|---|---|---|
| 1k | 178.8 | 8.5 | 29.5 GB | view |
| 4k | 129.0 | 6.3 | 32.5 GB | view |
| 8k | 112.1 | 4.8 | 34.4 GB | view |
| 16k | 58.0 | 3.6 | 37.4 GB | current |
Batching Results
| Batch Size | TG tok/s | Speedup |
|---|---|---|
| 1× | 8.5 | 1.00× |
| 2× | 9.2 | 1.08× |
| 4× | 13.1 | 1.54× |