← Back to community benchmarks
Qwen3.5-397B-A17B
Performance
4k
tokens
525.2
PP tok/s
35.0
TG tok/s
7799
TTFT (ms)
210.7
Peak mem (GB)
Hardware
Chip
M3 Ultra (80c)
Memory
512 GB
GPU Cores
80
Software
oMLX
v0.2.10
macOS
macOS 26.3.1
Context
4,096
Performance by Context Length
| Context | PP tok/s | TG tok/s | Peak Mem | |
|---|---|---|---|---|
| 1k | 492.6 | 36.5 | 210.4 GB | view |
| 4k | 525.2 | 35.0 | 210.7 GB | current |
| 8k | 519.9 | 33.2 | 211.0 GB | view |
Batching Results
| Batch Size | TG tok/s | Speedup |
|---|---|---|
| 1× | 36.5 | 1.00× |
| 2× | 57.8 | 1.58× |
| 4× | 81.8 | 2.24× |