← Back to community benchmarks
Qwen3.6-35B-A3B-Claude-4.7-Opus-Reasoning-Distilled-VLMoQ8
Performance
16k
tokens
611.1
PP tok/s
57.5
TG tok/s
26812
TTFT (ms)
37.4
Peak mem (GB)
Hardware
Chip
M2 Max (38c)
Memory
96 GB
GPU Cores
38
Software
oMLX
v0.3.8
macOS
macOS 26.4.1
Context
16,384
Performance by Context Length
| Context | PP tok/s | TG tok/s | Peak Mem | |
|---|---|---|---|---|
| 1k | 635.6 | 67.3 | 35.4 GB | view |
| 4k | 714.5 | 64.3 | 36.2 GB | view |
| 8k | 645.3 | 61.6 | 36.5 GB | view |
| 16k | 611.1 | 57.5 | 37.4 GB | current |
| 32k | 543.8 | 48.0 | 38.8 GB | view |
| 64k | 443.5 | 36.7 | 41.8 GB | view |
| 128k | 318.6 | 26.8 | 47.8 GB | view |
Batching Results
| Batch Size | TG tok/s | Speedup |
|---|---|---|
| 1× | 67.3 | 1.00× |
| 2× | 92.4 | 1.37× |
| 4× | 120.1 | 1.78× |
| 8× | 143.0 | 2.12× |