← Back to community benchmarks

Qwen3-VL-8B-Instruct

M5 Pro (16c) · 24 GB · 4bit · 2026-04-26
Performance
1k
tokens
1,241
PP tok/s
57.8
TG tok/s
825
TTFT (ms)
5.0
Peak mem (GB)
Hardware
Chip M5 Pro (16c)
Memory 24 GB
GPU Cores 16
Software
oMLX v0.3.7
macOS macOS 26.4.1
Context 1,024
Performance by Context Length
Context PP tok/s TG tok/s Peak Mem
1k 1,241 57.8 5.0 GB current
4k 1,355 50.3 5.5 GB view
Batching Results
Batch Size TG tok/s Speedup
57.8 1.00×
92.3 1.60×
113.2 1.96×