← Back to community benchmarks
Qwen3.5-35B-A3B
Performance
64k
tokens
483.3
PP tok/s
23.6
TG tok/s
135613
TTFT (ms)
24.9
Peak mem (GB)
Hardware
Chip
M5 (10c)
Memory
32 GB
GPU Cores
10
Software
oMLX
v0.3.5
macOS
macOS 26.4
Context
65,536
Performance by Context Length
| Context | PP tok/s | TG tok/s | Peak Mem | |
|---|---|---|---|---|
| 1k | 814.6 | 59.0 | 19.2 GB | view |
| 4k | 1,016 | 56.6 | 19.9 GB | view |
| 8k | 994.9 | 53.3 | 20.2 GB | view |
| 16k | 902.1 | 46.8 | 20.9 GB | view |
| 32k | 730.4 | 36.3 | 22.2 GB | view |
| 64k | 483.3 | 23.6 | 24.9 GB | current |
Batching Results
| Batch Size | TG tok/s | Speedup |
|---|---|---|
| 1× | 59.0 | 1.00× |
| 2× | 85.6 | 1.45× |
| 4× | 99.6 | 1.69× |