Configure a rig →
Qwen14B params128k contextpopular

Qwen 3 14B locally

Qwen 3 14B is an open-weight LLM from the Qwen family with 14B parameters. Main use: chat, RAG and general assistance. Detected minimum hardware: RTX 3080 10GB (10 GB).

Check on my GPU →Hugging Face ↗
Technical facts
Parameters14B
Max context128k
Q4_K_M8.8 GB
Q5_K_M10.8 GB
Q815.6 GB
FP1631.3 GB
FamilyQwen
Last sync2026-05-12

Available quantizations

Q4_K_M
8.8GB

Acceptable. Good compromise when VRAM is limited.

Q5_K_M
10.8GB

Good quality. Sweet spot for size and precision.

Q8
15.6GB

Near-FP16 quality. Comfortable for production.

FP16
31.3GB

Reference precision. Maximum quality, doubled VRAM.

Compatible GPUs

GPUs that can run Qwen 3 14B on a single card, ranked by VRAM headroom.

RTX 3080 10GB
NVIDIA
10 GB · RTX 30
8.8 / 10 GBtight · Q4
Radeon RX 6700
AMD
10 GB · RDNA 2
8.8 / 10 GBtight · Q4
Arc B570
Intel
10 GB · Arc Battlemage
8.8 / 10 GBtight · Q4
GTX 1080 Ti
NVIDIA
11 GB · GTX 10
8.8 / 11 GBcomfortable · Q4
RTX 2080 Ti
NVIDIA
11 GB · RTX 20
8.8 / 11 GBcomfortable · Q4
RTX 2060 12GB
NVIDIA
12 GB · RTX 20
8.8 / 12 GBcomfortable · Q4
RTX 3060 12GB
NVIDIA
12 GB · RTX 30
8.8 / 12 GBcomfortable · Q4
RTX 3080 12GB
NVIDIA
12 GB · RTX 30
8.8 / 12 GBcomfortable · Q4
RTX 3080 Ti
NVIDIA
12 GB · RTX 30
8.8 / 12 GBcomfortable · Q4
RTX 4070
NVIDIA
12 GB · RTX 40
8.8 / 12 GBcomfortable · Q4
RTX 4070 Super
NVIDIA
12 GB · RTX 40
8.8 / 12 GBcomfortable · Q4
RTX 4070 Ti
NVIDIA
12 GB · RTX 40
8.8 / 12 GBcomfortable · Q4

Recommended multi-GPU rigs

For Qwen 3 14B at higher quantization or with more context, a multi-GPU rig gives more headroom.

2× GTX 1060 6GB
NVIDIA
12 GB · GTX 10
8.8 / 12 GBcomfortable · Q4
2× GTX 1660
NVIDIA
12 GB · GTX 16
8.8 / 12 GBcomfortable · Q4
2× GTX 1660 Super
NVIDIA
12 GB · GTX 16
8.8 / 12 GBcomfortable · Q4
2× GTX 1660 Ti
NVIDIA
12 GB · GTX 16
8.8 / 12 GBcomfortable · Q4
2× RTX 2060 6GB
NVIDIA
12 GB · RTX 20
8.8 / 12 GBcomfortable · Q4
2× RTX 3050 6GB
NVIDIA
12 GB · RTX 30
8.8 / 12 GBcomfortable · Q4
2× Arc A380
Intel
12 GB · Arc Alchemist
8.8 / 12 GBcomfortable · Q4
2× GTX 1070
NVIDIA
16 GB · GTX 10
10.8 / 16 GBcomfortable · Q5

Recommended rig

2× GTX 1060 6GB

Qwen 3 14B 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