Qwen3 30B-A3B · Q4_K_M
GGUF · Q4_K_M
- Model-size signal
- 19 GB
- Minimum load
- 21.9 GB
- Planning memory
- 25.7 GB
- GPU VRAM
- 24 GB
24 GB dedicated VRAM meets the 21.9 GB minimum-load planning level, but not the 25.7 GB recommended headroom.
Expect less room for long context or concurrent GPU work.