← Back to community benchmarks
Qwen3.5-35B-A3B
Performance
32k
tokens
969.1
PP tok/s
44.3
TG tok/s
33814
TTFT (ms)
37.8
Peak mem (GB)
Hardware
Chip
M2 Ultra (60c)
Memory
64 GB
GPU Cores
60
Software
oMLX
v0.2.6
macOS
macOS 26.3
Context
32,768
Performance by Context Length
| Context | PP tok/s | TG tok/s | Peak Mem | |
|---|---|---|---|---|
| 1k | 1,025 | 67.3 | 36.2 GB | view |
| 4k | 1,123 | 64.8 | 36.3 GB | view |
| 8k | 1,120 | 60.2 | 36.5 GB | view |
| 16k | 1,068 | 54.5 | 37.0 GB | view |
| 32k | 969.1 | 44.3 | 37.8 GB | current |
| 64k | 801.8 | 32.7 | 39.5 GB | view |
Batching Results
| Batch Size | TG tok/s | Speedup |
|---|---|---|
| 1× | 67.3 | 1.00× |
| 2× | 114.9 | 1.71× |
| 4× | 182.2 | 2.71× |
| 8× | 263.8 | 3.92× |