← Back to community benchmarks
Qwen3.5-35B-A3B-Claude-4.6-Opus-Reasoning-Distilled
Performance
32k
tokens
954.1
PP tok/s
38.1
TG tok/s
34346
TTFT (ms)
20.8
Peak mem (GB)
Hardware
Chip
M2 Ultra (60c)
Memory
64 GB
GPU Cores
60
Software
oMLX
v0.2.9
macOS
macOS 15.7.3
Context
32,768
Performance by Context Length
| Context | PP tok/s | TG tok/s | Peak Mem | |
|---|---|---|---|---|
| 1k | 646.5 | 82.6 | 19.2 GB | view |
| 4k | 1,005 | 79.3 | 19.4 GB | view |
| 8k | 1,078 | 72.8 | 19.6 GB | view |
| 16k | 1,042 | 59.4 | 20.0 GB | view |
| 32k | 954.1 | 38.1 | 20.8 GB | current |
| 64k | 789.0 | 23.6 | 22.5 GB | view |
Batching Results
| Batch Size | TG tok/s | Speedup |
|---|---|---|
| 1× | 82.6 | 1.00× |
| 2× | 137.7 | 1.67× |
| 4× | 191.1 | 2.31× |