← Back to community benchmarks
Qwen3.5-397B-A17B9bit
Performance
4k
tokens
488.0
PP tok/s
25.7
TG tok/s
8394
TTFT (ms)
417.4
Peak mem (GB)
Hardware
Chip
M3 Ultra (80c)
Memory
512 GB
GPU Cores
80
Software
oMLX
v0.2.12
macOS
macOS 26.3.1
Context
4,096
Performance by Context Length
| Context | PP tok/s | TG tok/s | Peak Mem | |
|---|---|---|---|---|
| 1k | 424.3 | 26.4 | 417.1 GB | view |
| 4k | 488.0 | 25.7 | 417.4 GB | current |
| 8k | 490.1 | 23.7 | 417.8 GB | view |
Batching Results
| Batch Size | TG tok/s | Speedup |
|---|---|---|
| 1× | 26.4 | 1.00× |
| 2× | 42.0 | 1.59× |
| 4× | 55.6 | 2.11× |