← Back to community benchmarks
Qwen3.6-35B-A3B-Holo3-Qwopus-AgentWorld-qx64-hi
Performance
16k
tokens
1,273
PP tok/s
79.2
TG tok/s
12875
TTFT (ms)
30.3
Peak mem (GB)
Hardware
Chip
M4 Max (40c)
Memory
128 GB
GPU Cores
40
Software
oMLX
v0.5.0
macOS
macOS 26.5.2
Context
16,384
Performance by Context Length
| Context | PP tok/s | TG tok/s | Peak Mem | |
|---|---|---|---|---|
| 1k | 1,367 | 90.9 | 29.1 GB | view |
| 4k | 1,692 | 90.0 | 29.8 GB | view |
| 8k | 1,554 | 87.6 | 30.0 GB | view |
| 16k | 1,273 | 79.2 | 30.3 GB | current |
| 32k | 888.8 | 66.1 | 30.9 GB | view |
| 64k | 217.4 | 19.6 | 32.7 GB | view |
Batching Results
| Batch Size | TG tok/s | Speedup |
|---|---|---|
| 1× | 90.9 | 1.00× |
| 2× | 61.5 | 0.68× |
| 4× | 101.4 | 1.12× |
| 8× | 132.5 | 1.46× |