← Back to community benchmarks
Qwen3.5-35B-A3B
Performance
8k
tokens
1,103
PP tok/s
72.1
TG tok/s
7425
TTFT (ms)
20.4
Peak mem (GB)
Hardware
Chip
M2 Ultra (60c)
Memory
64 GB
GPU Cores
60
Software
oMLX
v0.2.6
macOS
macOS 26.3
Context
8,192
Performance by Context Length
| Context | PP tok/s | TG tok/s | Peak Mem | |
|---|---|---|---|---|
| 1k | 1,039 | 76.3 | 20.0 GB | view |
| 4k | 1,139 | 73.5 | 20.2 GB | view |
| 8k | 1,103 | 72.1 | 20.4 GB | current |
| 16k | 1,075 | 62.1 | 20.8 GB | view |
| 32k | 977.7 | 51.8 | 21.6 GB | view |
| 64k | 810.1 | 37.9 | 23.3 GB | view |
Batching Results
| Batch Size | TG tok/s | Speedup |
|---|---|---|
| 1× | 76.3 | 1.00× |
| 2× | 133.9 | 1.75× |
| 4× | 192.9 | 2.53× |
| 8× | 252.4 | 3.31× |