← Back to community benchmarks
Qwen3.6-35B-A3B-Claude-4.7-Opus-abliterated-hermes-oQ4-fp16-mtp
Performance
16k
tokens
1,843
PP tok/s
78.1
TG tok/s
8888
TTFT (ms)
21.9
Peak mem (GB)
Hardware
Chip
M2 Ultra (60c)
Memory
64 GB
GPU Cores
60
Software
oMLX
v0.5.1
macOS
macOS 26.5.2
Context
16,384
Performance by Context Length
| Context | PP tok/s | TG tok/s | Peak Mem | |
|---|---|---|---|---|
| 1k | 1,944 | 85.5 | 19.9 GB | view |
| 4k | 2,055 | 84.5 | 20.7 GB | view |
| 8k | 1,995 | 82.7 | 21.1 GB | view |
| 16k | 1,843 | 78.1 | 21.9 GB | current |
| 32k | 1,573 | 70.6 | 23.7 GB | view |
| 64k | 1,139 | 62.5 | 26.0 GB | view |
Batching Results
| Batch Size | TG tok/s | Speedup |
|---|---|---|
| 1× | 85.5 | 1.00× |
| 2× | 142.5 | 1.67× |
| 4× | 365.6 | 4.28× |