gpu 12gb llm-inference

RTX 3060 12GB

Updated Sep 6, 2026

VRAM

12 GB

Bandwidth

360 GB/s

TDP

170 W

MSRP

$280

What can I run VRAM calculator

Operator notes for RTX 3060 12GB are still landing. Specs and model-fit math below are live from directory data.

Models that fit on this card

Best quant that still fits the listed VRAM, from published listings only.

d1-omni-600M

0.6B

1.3/12 GB

FP16

Tev1-0.8B-experimental

0.9B

2/12 GB

FP16

Qwen3-0.6B

0.8B

3.8/12 GB

FP16

Qwen2.5-1.5B-Instruct

1.5B

4.2/12 GB

FP16

Qwen2.5-3B-Instruct

3.1B

7.7/12 GB

FP16

Qwen3-4B

4.0B

8/12 GB

FP8

gemma-4-E4B

8.0B

9.7/12 GB

FP8

Qwen2.5-7B-Instruct

7.6B

10.3/12 GB

FP8

Tev1-4B-experimental

4.7B

10.4/12 GB

FP16

Qwen3.6-35B-A3B-NVFP4

19B

10.5/12 GB

Q3_K_M

Qwen3-8B

8.2B

11.6/12 GB

Q5_K_M

Close but over budget

These need more VRAM than this card lists. Shown so you do not chase a false fit.

gpt-oss-20b

21B · needs ~12.059 GB

12.1/12 GB

Q2_K

OTel-2.0-LLM-31B-IT

31B · needs ~14.635 GB

14.6/12 GB

Q2_K

AdvancedMathBench-AutoVerifier

36B · needs ~16.831 GB

16.8/12 GB

Q2_K