← Back to community benchmarks
Qwen3.5-27B-Claude-4.6-Opus-Distilled
Performance
4k
tokens
409.5
PP tok/s
26.1
TG tok/s
10002
TTFT (ms)
23.6
Peak mem (GB)
Hardware
Chip
M3 Ultra (80c)
Memory
512 GB
GPU Cores
80
Software
oMLX
v0.2.19
macOS
macOS 26.2
Context
4,096
Performance by Context Length
| Context | PP tok/s | TG tok/s | Peak Mem | |
|---|---|---|---|---|
| 1k | 384.0 | 27.0 | 22.1 GB | view |
| 4k | 409.5 | 26.1 | 23.6 GB | current |
| 8k | 411.3 | 25.2 | 24.3 GB | view |
| 16k | 400.3 | 23.6 | 25.6 GB | view |
| 32k | 376.4 | 21.0 | 28.1 GB | view |
| 64k | 327.3 | 17.1 | 33.1 GB | view |
Batching Results
| Batch Size | TG tok/s | Speedup |
|---|---|---|
| 1× | 27.0 | 1.00× |
| 2× | 46.4 | 1.72× |
| 4× | 55.2 | 2.04× |
| 8× | 60.6 | 2.24× |