← Back to community benchmarks
Qwen3.5-27B-Claude-4.6-Opus-Distilled
Performance
128k
tokens
82.2
PP tok/s
3.6
TG tok/s
1593623
TTFT (ms)
33.0
Peak mem (GB)
Hardware
Chip
M3 Max (40c)
Memory
128 GB
GPU Cores
40
Software
oMLX
v0.2.13
macOS
macOS 26.4
Context
131,072
Performance by Context Length
| Context | PP tok/s | TG tok/s | Peak Mem | |
|---|---|---|---|---|
| 1k | 204.5 | 22.3 | 15.8 GB | view |
| 4k | 193.8 | 21.0 | 16.1 GB | view |
| 8k | 181.6 | 19.0 | 16.6 GB | view |
| 16k | 154.7 | 9.6 | 17.5 GB | view |
| 32k | 102.5 | 9.6 | 19.2 GB | view |
| 64k | 104.3 | 5.2 | 22.8 GB | view |
| 128k | 82.2 | 3.6 | 33.0 GB | current |
| 195k | 60.6 | 2.4 | 40.6 GB | view |
Batching Results
| Batch Size | TG tok/s | Speedup |
|---|---|---|
| 1× | 22.3 | 1.00× |
| 2× | 16.7 | 0.75× |
| 4× | 21.9 | 0.98× |