← Back to community benchmarks
Qwen3.5-27B-Claude-4.6-OS-INSTRUCT-mxfp8
Performance
32k
tokens
210.5
PP tok/s
10.4
TG tok/s
155663
TTFT (ms)
31.8
Peak mem (GB)
Hardware
Chip
M4 Max (40c)
Memory
128 GB
GPU Cores
40
Software
oMLX
v0.2.7
macOS
macOS 26.3.1
Context
32,768
Performance by Context Length
| Context | PP tok/s | TG tok/s | Peak Mem | |
|---|---|---|---|---|
| 1k | 228.6 | 15.5 | 28.4 GB | view |
| 4k | 231.6 | 15.0 | 28.7 GB | view |
| 8k | 230.3 | 13.8 | 29.2 GB | view |
| 16k | 223.4 | 12.3 | 30.0 GB | view |
| 32k | 210.5 | 10.4 | 31.8 GB | current |
| 64k | 184.5 | 7.6 | 35.4 GB | view |
Batching Results
| Batch Size | TG tok/s | Speedup |
|---|---|---|
| 1× | 15.5 | 1.00× |
| 2× | 24.2 | 1.56× |
| 4× | 23.5 | 1.52× |