← Back to community benchmarks
Qwen3.5-REAP-97B-A10B
Performance
16k
tokens
164.3
PP tok/s
27.9
TG tok/s
99710
TTFT (ms)
55.1
Peak mem (GB)
Hardware
Chip
M1 Max (32c)
Memory
64 GB
GPU Cores
32
Software
oMLX
v0.5.2rc1
macOS
macOS 26.5.2
Context
16,384
Performance by Context Length
| Context | PP tok/s | TG tok/s | Peak Mem | |
|---|---|---|---|---|
| 1k | 180.3 | 30.5 | 52.4 GB | view |
| 16k | 164.3 | 27.9 | 55.1 GB | current |
| 64k | 104.6 | 18.8 | 56.8 GB | view |
Batching Results
| Batch Size | TG tok/s | Speedup |
|---|---|---|
| 1× | 30.5 | 1.00× |
| 2× | 37.6 | 1.23× |
| 4× | 75.6 | 2.48× |