← Back to community benchmarks
Qwen3.5-9B-Claude-4.6-HighIQ-INSTRUCT-HERETIC-UNCENSORED
Performance
4k
tokens
1,389
PP tok/s
103.6
TG tok/s
2949
TTFT (ms)
6.5
Peak mem (GB)
Hardware
Chip
M3 Ultra (80c)
Memory
512 GB
GPU Cores
80
Software
oMLX
v0.3.7
macOS
macOS 26.4.1
Context
4,096
Performance by Context Length
| Context | PP tok/s | TG tok/s | Peak Mem | |
|---|---|---|---|---|
| 1k | 1,115 | 107.0 | 5.8 GB | view |
| 4k | 1,389 | 103.6 | 6.5 GB | current |
Batching Results
| Batch Size | TG tok/s | Speedup |
|---|---|---|
| 1× | 107.0 | 1.00× |
| 2× | 183.2 | 1.71× |
| 4× | 246.3 | 2.30× |
| 8× | 291.6 | 2.73× |