← Back to community benchmarks
Qwen3-6-35B-A3B-oQ8
Performance
8k
tokens
1,691
PP tok/s
80.1
TG tok/s
4845
TTFT (ms)
36.5
Peak mem (GB)
Hardware
Chip
M4 Max (40c)
Memory
128 GB
GPU Cores
40
Software
oMLX
v0.3.6
macOS
macOS 15.7.4
Context
8,192
Performance by Context Length
| Context | PP tok/s | TG tok/s | Peak Mem | |
|---|---|---|---|---|
| 1k | 1,349 | 86.7 | 35.4 GB | view |
| 4k | 1,696 | 84.1 | 36.2 GB | view |
| 8k | 1,691 | 80.1 | 36.5 GB | current |
| 16k | 1,586 | 74.1 | 37.1 GB | view |
| 32k | 1,375 | 65.5 | 38.5 GB | view |
| 64k | 1,066 | 51.9 | 41.2 GB | view |
Batching Results
| Batch Size | TG tok/s | Speedup |
|---|---|---|
| 1× | 86.7 | 1.00× |
| 2× | 140.2 | 1.62× |
| 4× | 200.0 | 2.31× |
| 8× | 259.1 | 2.99× |