← Back to community benchmarks
Qwen3.5-35B-A3B
Performance
16k
tokens
614.5
PP tok/s
44.1
TG tok/s
26661
TTFT (ms)
20.8
Peak mem (GB)
Hardware
Chip
M3 Pro (18c)
Memory
36 GB
GPU Cores
18
Software
oMLX
v0.2.13
macOS
macOS 26.3
Context
16,384
Performance by Context Length
| Context | PP tok/s | TG tok/s | Peak Mem | |
|---|---|---|---|---|
| 1k | 648.3 | 56.7 | 20.0 GB | view |
| 4k | 680.9 | 54.1 | 20.2 GB | view |
| 8k | 670.6 | 50.5 | 20.4 GB | view |
| 16k | 614.5 | 44.1 | 20.8 GB | current |
| 32k | 527.3 | 35.6 | 21.6 GB | view |
| 64k | 401.9 | 23.9 | 23.3 GB | view |
Batching Results
| Batch Size | TG tok/s | Speedup |
|---|---|---|
| 1× | 56.7 | 1.00× |
| 2× | 89.1 | 1.57× |
| 4× | 118.0 | 2.08× |