← Back to community benchmarks
Qwen:Qwen2.5-VL-7B-Instruct
Performance
16k
tokens
634.3
PP tok/s
23.5
TG tok/s
25832
TTFT (ms)
15.6
Peak mem (GB)
Hardware
Chip
M5 Max
Memory
36 GB
Software
oMLX
v0.3.8rc1
macOS
macOS 26.4.1
Context
16,384
Performance by Context Length
| Context | PP tok/s | TG tok/s | Peak Mem | |
|---|---|---|---|---|
| 1k | 539.3 | 27.5 | 14.7 GB | view |
| 4k | 604.3 | 26.2 | 15.0 GB | view |
| 8k | 668.5 | 25.5 | 15.2 GB | view |
| 16k | 634.3 | 23.5 | 15.6 GB | current |
Batching Results
| Batch Size | TG tok/s | Speedup |
|---|---|---|
| 1× | 27.5 | 1.00× |
| 2× | 28.1 | 1.02× |
| 4× | 55.0 | 2.00× |
| 8× | 103.7 | 3.77× |