← Back to community benchmarks
Qwen3.5-27B-Claude-4.6-Opus-Reasoning-Distilled-v2
Performance
16k
tokens
404.3
PP tok/s
19.0
TG tok/s
40525
TTFT (ms)
31.7
Peak mem (GB)
Hardware
Chip
M3 Ultra (80c)
Memory
512 GB
GPU Cores
80
Software
oMLX
v0.3.4
macOS
macOS 15.4.1
Context
16,384
Performance by Context Length
| Context | PP tok/s | TG tok/s | Peak Mem | |
|---|---|---|---|---|
| 1k | 379.6 | 21.7 | 28.3 GB | view |
| 4k | 415.1 | 21.1 | 29.8 GB | view |
| 8k | 414.6 | 20.3 | 30.4 GB | view |
| 16k | 404.3 | 19.0 | 31.7 GB | current |
Batching Results
| Batch Size | TG tok/s | Speedup |
|---|---|---|
| 1× | 21.7 | 1.00× |
| 2× | 38.5 | 1.77× |
| 4× | 57.2 | 2.64× |
| 8× | 4.2 | 0.19× |