← Back to community benchmarks
Qwen3.5-35B-A3B-Hereticmxfp8
Performance
16k
tokens
776.3
PP tok/s
48.6
TG tok/s
21106
TTFT (ms)
35.1
Peak mem (GB)
Hardware
Chip
M4 Pro (20c)
Memory
64 GB
GPU Cores
20
Software
oMLX
v0.2.8
macOS
macOS 26.3
Context
16,384
Performance by Context Length
| Context | PP tok/s | TG tok/s | Peak Mem | |
|---|---|---|---|---|
| 1k | 749.0 | 58.5 | 34.4 GB | view |
| 4k | 842.7 | 56.3 | 34.5 GB | view |
| 8k | 830.3 | 53.5 | 34.7 GB | view |
| 16k | 776.3 | 48.6 | 35.1 GB | current |
| 32k | 647.9 | 41.4 | 36.0 GB | view |
| 64k | 514.5 | 28.5 | 37.6 GB | view |
Batching Results
| Batch Size | TG tok/s | Speedup |
|---|---|---|
| 1× | 58.5 | 1.00× |
| 2× | 81.4 | 1.39× |
| 4× | 94.1 | 1.61× |
| 8× | 106.8 | 1.83× |