Configure a rig →
Qwen30B paramspopular

Qwen3 30B A3B locally

Qwen3 30B A3B is an open-weight LLM from the Qwen family with 30B parameters. Main use: chat, RAG and general assistance. Detected minimum hardware: RTX A4500 (20 GB).

Check on my GPU →Hugging Face ↗
Technical facts
Parameters30B
Q4_K_M18.9 GB
Q5_K_M23.1 GB
Q833.5 GB
FP1667.1 GB
FamilyQwen
Last sync2026-05-12

Available quantizations

Q4_K_M
18.9GB

Acceptable. Good compromise when VRAM is limited.

Q5_K_M
23.1GB

Good quality. Sweet spot for size and precision.

Q8
33.5GB

Near-FP16 quality. Comfortable for production.

FP16
67.1GB

Reference precision. Maximum quality, doubled VRAM.

Compatible GPUs

GPUs that can run Qwen3 30B A3B on a single card, ranked by VRAM headroom.

RTX A4500
NVIDIA
20 GB · RTX A (Ampere)
18.9 / 20 GBtight · Q4
RTX 4000 Ada
NVIDIA
20 GB · RTX Ada
18.9 / 20 GBtight · Q4
Radeon RX 7900 XT
AMD
20 GB · RDNA 3
18.9 / 20 GBtight · Q4
TITAN RTX
NVIDIA
24 GB · RTX 20
18.9 / 24 GBcomfortable · Q4
RTX 3090
NVIDIA
24 GB · RTX 30
18.9 / 24 GBcomfortable · Q4
RTX 3090 Ti
NVIDIA
24 GB · RTX 30
18.9 / 24 GBcomfortable · Q4
RTX 4090
NVIDIA
24 GB · RTX 40
18.9 / 24 GBcomfortable · Q4
Quadro RTX 6000
NVIDIA
24 GB · Quadro RTX
18.9 / 24 GBcomfortable · Q4
RTX A5000
NVIDIA
24 GB · RTX A (Ampere)
18.9 / 24 GBcomfortable · Q4
RTX A5500
NVIDIA
24 GB · RTX A (Ampere)
18.9 / 24 GBcomfortable · Q4
RTX 4500 Ada
NVIDIA
24 GB · RTX Ada
18.9 / 24 GBcomfortable · Q4
NVIDIA A10
NVIDIA
24 GB · Ampere DC
18.9 / 24 GBcomfortable · Q4

Recommended multi-GPU rigs

For Qwen3 30B A3B at higher quantization or with more context, a multi-GPU rig gives more headroom.

2× RTX 3080 10GB
NVIDIA
20 GB · RTX 30
18.9 / 20 GBtight · Q4
2× Radeon RX 6700
AMD
20 GB · RDNA 2
18.9 / 20 GBtight · Q4
2× Arc B570
Intel
20 GB · Arc Battlemage
18.9 / 20 GBtight · Q4
2× GTX 1080 Ti
NVIDIA
22 GB · GTX 10
18.9 / 22 GBtight · Q4
2× RTX 2080 Ti
NVIDIA
22 GB · RTX 20
18.9 / 22 GBtight · Q4
4× GTX 1060 6GB
NVIDIA
24 GB · GTX 10
18.9 / 24 GBcomfortable · Q4
4× GTX 1660
NVIDIA
24 GB · GTX 16
18.9 / 24 GBcomfortable · Q4
4× GTX 1660 Super
NVIDIA
24 GB · GTX 16
18.9 / 24 GBcomfortable · Q4

Recommended rig

4× GTX 1060 6GB

Qwen3 30B A3B with Ubuntu, vLLM, Open WebUI and the model already downloaded.

Configure →

Similar models

VRAM estimates: parameters x bits/8 plus margin. Real performance varies by engine, context length and batch size.
sync: 2026-05-12