← Back to community benchmarks
Qwen3.5-35B-A3BMXFP4
Performance
16k
tokens
696.4
PP tok/s
47.1
TG tok/s
23527
TTFT (ms)
22.8
Peak mem (GB)
Hardware
Chip
M1 Max (32c)
Memory
64 GB
GPU Cores
32
Software
oMLX
v0.3.6
macOS
macOS 26.4
Context
16,384
Performance by Context Length
| Context | PP tok/s | TG tok/s | Peak Mem | |
|---|---|---|---|---|
| 1k | 629.5 | 66.3 | 19.2 GB | view |
| 4k | 737.7 | 62.9 | 20.8 GB | view |
| 8k | 737.2 | 57.2 | 21.5 GB | view |
| 16k | 696.4 | 47.1 | 22.8 GB | current |
| 32k | 540.0 | 34.0 | 25.4 GB | view |
Batching Results
| Batch Size | TG tok/s | Speedup |
|---|---|---|
| 1× | 66.3 | 1.00× |
| 2× | 93.8 | 1.41× |