← Back to community benchmarks
Qwen3.5-35B-A3B
Performance
1k
tokens
2,643
PP tok/s
96.8
TG tok/s
388
TTFT (ms)
20.1
Peak mem (GB)
Hardware
Chip
M3 Ultra (80c)
Memory
512 GB
GPU Cores
80
Software
oMLX
v0.2.10
macOS
macOS 26.2
Context
1,024
Performance by Context Length
| Context | PP tok/s | TG tok/s | Peak Mem | |
|---|---|---|---|---|
| 1k | 2,643 | 96.8 | 20.1 GB | current |
| 4k | 2,757 | 93.2 | 20.2 GB | view |
| 8k | 2,672 | 89.0 | 20.4 GB | view |
| 16k | 2,468 | 81.8 | 20.8 GB | view |
| 32k | 2,130 | 70.7 | 21.6 GB | view |
| 64k | 1,627 | 53.8 | 23.3 GB | view |
Batching Results
| Batch Size | TG tok/s | Speedup |
|---|---|---|
| 1× | 96.8 | 1.00× |
| 2× | 172.2 | 1.78× |
| 4× | 283.5 | 2.93× |
| 8× | 428.2 | 4.42× |