Configure a rig →

Qwen3B params128k context

Qwen 2.5 VL 3B locally

Qwen 2.5 VL 3B is an open-weight LLM from the Qwen family with 3B parameters. Main use: multimodal image workflows. Detected minimum hardware: GTX 1650 (4 GB).

Check on my GPU →Hugging Face ↗

Technical facts

Parameters3B

Max context128k

Q4_K_M1.9 GB

Q5_K_M2.3 GB

Q83.4 GB

FP166.7 GB

FamilyQwen

Last sync2026-05-12

Available quantizations

GGUF weights

Q4_K_M

1.9GB

Acceptable. Good compromise when VRAM is limited.

Q5_K_M

2.3GB

Good quality. Sweet spot for size and precision.

Q8

3.4GB

Near-FP16 quality. Comfortable for production.

FP16

6.7GB

Reference precision. Maximum quality, doubled VRAM.

Compatible GPUs

12 single-GPU

GPUs that can run Qwen 2.5 VL 3B on a single card, ranked by VRAM headroom.

3.4 / 4 GBcomfortable · Q8

3.4 / 6 GBcomfortable · Q8

3.4 / 6 GBcomfortable · Q8

3.4 / 6 GBcomfortable · Q8

3.4 / 6 GBcomfortable · Q8

3.4 / 6 GBcomfortable · Q8

3.4 / 6 GBcomfortable · Q8

6 GB · Arc Alchemist

3.4 / 6 GBcomfortable · Q8

6.7 / 8 GBcomfortable · FP16

6.7 / 8 GBcomfortable · FP16

6.7 / 8 GBcomfortable · FP16

6.7 / 8 GBcomfortable · FP16

Recommended multi-GPU rigs

2x / 4x consumer GPUs

For Qwen 2.5 VL 3B at higher quantization or with more context, a multi-GPU rig gives more headroom.

6.7 / 8 GBcomfortable · FP16

2× GTX 1060 6GB

12 GB · GTX 10

6.7 / 12 GBcomfortable · FP16

12 GB · GTX 16

6.7 / 12 GBcomfortable · FP16

2× GTX 1660 Super

12 GB · GTX 16

6.7 / 12 GBcomfortable · FP16

2× GTX 1660 Ti

12 GB · GTX 16

6.7 / 12 GBcomfortable · FP16

2× RTX 2060 6GB

12 GB · RTX 20

6.7 / 12 GBcomfortable · FP16

2× RTX 3050 6GB

12 GB · RTX 30

6.7 / 12 GBcomfortable · FP16

12 GB · Arc Alchemist

6.7 / 12 GBcomfortable · FP16

Recommended rig

2× GTX 1650

Qwen 2.5 VL 3B with Ubuntu, vLLM, Open WebUI and the model already downloaded.

Similar models

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