← Back to community benchmarks
Qwen3.6-27B-Claude-Opus-Reasoning-Distill-v2-abliterated-8bit
Performance
1k
tokens
173.2
PP tok/s
20.1
TG tok/s
5913
TTFT (ms)
28.3
Peak mem (GB)
Hardware
Chip
M2 Ultra (60c)
Memory
64 GB
GPU Cores
60
Software
oMLX
v0.4.1
macOS
macOS 26.2
Context
1,024
Performance by Context Length
| Context | PP tok/s | TG tok/s | Peak Mem | |
|---|---|---|---|---|
| 1k | 173.2 | 20.1 | 28.3 GB | current |
| 4k | 176.9 | 19.7 | 29.7 GB | view |
Batching Results
| Batch Size | TG tok/s | Speedup |
|---|---|---|
| 1× | 20.1 | 1.00× |
| 2× | 32.9 | 1.64× |
| 4× | 42.7 | 2.12× |