Sentdex corrects himself: 9GB memory serves the default 4B Qwen model

Sentdex · x · 2026-09-22

Sentdex corrects his earlier statement: the 9GB of memory corresponds to the default 4-billion-parameter Qwen model, not an 8B one. He has also tried the 8B Qwen and other models, but says he'd stick with the 4B for now. Context: he was asked whether a local MLX stack on M5+ Macs could approach Jev's current benchmarks within 30 days.

Related event: Sentdex: running the default Qwen model locally takes about 9GB VRAM(3 posts)→

Original post →

More from Models

Models channel →