RunsOnMy ModelsGPUs & MacsBy VRAMBest GPUCan I run…?OllamaStart hereAPI pricesVRAM calculatorNew

What AI models run on the RTX 3060 12GB?

The RTX 3060 12GB has 12 GB of VRAM (360 GB/s). About 11.4 GB is usable for a model. 173 of 298 open models we track fit at 8K context.

173models fit (8K ctx)
154still fit at 32K ctx
31Blargest that fits
360 GB/smemory bandwidth

This card has 12 GB of VRAM. See every model that fits in 12 GB VRAM. Other 12 GB hardware: RTX 3080 12GB, RTX 4070, RTX 4070 Super, RTX 4070 Ti, RTX 5070, RX 7700 XT, Arc B580.

Compare for local LLMs: RTX 3060 12GB vs RTX 4060RTX 3060 12GB vs RTX 4060 Ti 16GBRTX 3060 12GB vs RTX 5060 Ti 16GBRTX 3060 12GB vs Arc B580 all comparisons

Using a specific tool? Ollama models that run on the RTX 3060 12GB · LM Studio · llama.cpp offload

Models that fit, largest first

ModelParamsBest quant (8K)Needs32K contextMax tok/s (est.)Details
GLM 4.7 Flash31B (4.6B act.)Q2_K11.4 GBno221details
Muse Glimmer 30B30BQ2_K11.4 GBno32details
Raon 1 Mini Preview29B (4.3B act.)Q2_K11.2 GBno222details
Arete 27B27BQ2_K11.2 GBno34details
MiniMax H3 Image PE27BQ2_K11.3 GBno34details
Trinity Mini26B (2.0B act.)Q2_K9.8 GBQ2_K508details
GLiDE 27B26BQ2_K10.5 GBno37details
DiffusionGemma 26B A4B IT26B (4.0B act.)Q2_K10.4 GBQ2_K240details
LFM2.5 24B A2B24B (2.0B act.)Q2_K9.2 GBQ2_K480details
ERNIE 4.5 21B A3B PT22BQ2_K8.6 GBQ2_K44details
ERNIE 4.5 21B A3B Thinking22BQ2_K8.6 GBQ2_K44details
Kimi VL A3B Thinking 250616B (3.0B act.)Q3_K_M8.4 GBQ3_K_M246details
Kimi VL A3B Thinking16B (3.0B act.)Q3_K_M8.4 GBQ3_K_M246details
Kimi VL A3B Instruct16B (3.0B act.)Q3_K_M8.7 GBQ3_K_M237details
DeepSeek MoE 16B Base16B (2.8B act.)Q3_K_M10.2 GBno246details
LLaDA2.1 Mini16B (1.4B act.)Q3_K_M8.4 GBQ3_K_M514details
Moonlight 16B A3B Instruct16B (3.0B act.)Q3_K_M8.7 GBQ3_K_M231details
DeepSeek Coder V2 Lite Instruct16B (2.7B act.)Q3_K_M8.0 GBQ3_K_M277details
DeepSeek V2 Lite16B (2.7B act.)Q4_K_M9.8 GBQ3_K_M221details
DeepSeek V2 Lite Chat16B (2.7B act.)Q3_K_M8.4 GBQ3_K_M261details
Qwen2.5 Coder 14B Instruct15BQ3_K_M9.0 GBno49details
DeepSeek R1 Distill Qwen 14B15BQ3_K_M9.0 GBno49details
Qwen3 14B15BQ3_K_M8.7 GBQ2_K50details
Phi 415BQ3_K_M9.0 GBno49details
Gemma 3 12B IT12BQ5_K_M9.6 GBQ4_K_M43details
Llama Guard 4 12B12BQ5_K_M10.2 GBQ2_K42details
KORMo Web 11B11BQ5_K_M9.3 GBQ2_K45details
Llama 3.2 11B Vision Instruct11BQ6_K10.0 GBQ2_K41details
Llama 3.2 11B Vision11BQ6_K10.0 GBQ2_K41details
GLM 4.6V Flash10BQ6_K8.7 GBQ6_K44details
GLM 4.1V 9B Thinking10BQ6_K8.7 GBQ6_K44details
Step3 VL 10B10BQ8_09.9 GBQ4_K_M41details
SenseNova Loop 9B Preview9.7BQ8_010.0 GBQ6_K38details
Qwen3.5 9B9.7BQ8_010.3 GBQ6_K37details
Ornith 1.5 9B9.7BQ8_010.0 GBQ6_K38details
SenseNova Loop 9B9.4BQ8_010.0 GBQ6_K38details
MiMo V2.6 Distill Qwen 9B9.4BQ8_010.0 GBQ6_K38details
GLM 4 9B9.4BQ6_K8.7 GBQ6_K44details
Apertus v1.5 8B8.9BQ6_K8.4 GBQ4_K_M49details
InternLM3 8B Instruct8.8BQ8_09.7 GBQ6_K38details
Granite 4.2 8B8.8BQ6_K8.6 GBQ3_K_M50details
Granite 4.1 8B8.8BQ6_K8.6 GBQ3_K_M50details
Granite 4.1 8B Llmon Preview8.8BQ6_K8.6 GBQ3_K_M50details
Qwen3 VL 8B Instruct8.8BQ8_09.9 GBQ4_K_M41details
Molmo2 8B8.7BQ8_09.9 GBQ4_K_M41details
LFM2.5 8B A1B8.5B (1.0B act.)Q8_09.1 GBQ8_0338details
Qwen2.5 VL 7B Instruct8.3BQ8_08.7 GBQ8_044details
UI TARS 1.5 7B8.3BQ8_08.7 GBQ8_044details
olmOCR 2 7B 10258.3BQ8_08.7 GBQ8_044details
olmOCR 7B 0225 Preview8.3BQ8_08.7 GBQ8_044details
UI TARS 7B DPO8.3BQ8_08.7 GBQ8_044details
Qwen3 8B8.2BQ8_09.9 GBQ4_K_M41details
DeepSeek R1 0528 Qwen3 8B8.2BQ8_09.9 GBQ4_K_M41details
Granite 3.3 8B Instruct8.2BQ8_09.8 GBQ4_K_M41details
Granite 3.1 8B Instruct8.2BQ8_09.8 GBQ4_K_M41details
Granite 3.2 8B Instruct8.2BQ8_09.8 GBQ4_K_M41details
Apertus 8B Instruct 25098.1BQ8_09.6 GBQ5_K_M42details
Llama 3.1 8B Instruct8.0BQ8_09.6 GBQ5_K_M42details
Llama 3.1 8B8.0BQ8_09.6 GBQ5_K_M42details
Hermes 3 Llama 3.1 8B8.0BQ8_09.6 GBQ5_K_M42details
Llama Guard 3 8B8.0BQ8_09.6 GBQ5_K_M42details
DeepSeek R1 Distill Llama 8B8.0BQ8_09.6 GBQ5_K_M42details
Meta Llama 3 8B Instruct8.0BQ8_09.6 GBQ5_K_M42details
Meta Llama Guard 2 8B8.0BQ8_09.6 GBQ5_K_M42details
Meta Llama 3.1 8B Instruct8.0BQ8_09.6 GBQ5_K_M42details
Aya Expanse 8B8.0BQ8_09.8 GBQ4_K_M42details
Ling 3.0 Tiny7.9B (1.4B act.)Q8_08.7 GBQ8_0245details
MiMo 7B RL7.8BQ8_09.4 GBQ4_K_M44details
InternLM2.5 7B Chat7.7BQ8_09.2 GBQ5_K_M44details
InternLM2 Chat 7B7.7BQ8_09.2 GBQ5_K_M44details
Qwen2.5 7B Instruct7.6BQ8_08.7 GBQ8_044details
DeepSeek R1 Distill Qwen 7B7.6BQ8_08.7 GBQ8_044details
Qwen2.5 Coder 7B Instruct7.6BQ8_08.7 GBQ8_044details
Olmo 3 7B Instruct7.3BQ6_K8.7 GBQ4_K_M60details
Olmo 3 1025 7B7.3BQ6_K8.7 GBQ4_K_M60details
Olmo 3 7B Think7.3BQ6_K8.7 GBQ4_K_M60details
Mistral 7B v0.17.2BQ8_08.6 GBQ6_K47details
Mistral 7B Instruct v0.27.2BQ8_08.6 GBQ6_K47details
Mistral 7B Instruct v0.17.2BQ8_08.6 GBQ6_K47details
Falcon 7B7.2BQ8_07.7 GBQ8_047details
Falcon 7B Instruct7.2BQ8_07.4 GBQ8_049details
Granite 4.0 H Tiny6.9B (1.5B act.)Q8_07.5 GBQ8_0231details
OLMoE 1B 7B 09246.9B (1.3B act.)Q8_08.4 GBQ6_K264details
DeepSeek Coder 7B Instruct v1.56.9BQ6_K9.6 GBno63details
DeepSeek Coder 6.7B Instruct6.7BQ6_K9.6 GBno65details
DeepSeek Coder 6.7B Base6.7BQ6_K9.6 GBno65details
Llama 2 7B HF6.7BQ6_K9.6 GBno65details
Yi 6B 200K6.1BQ8_07.0 GBQ8_056details
Yi 6B6.1BQ8_07.0 GBQ8_056details
Molmo2 4B4.9BF1610.0 GBQ8_041details
WeVisOne VL 4B4.8BF1610.0 GBQ8_041details
Mage VL Cowork4.7BF1610.0 GBQ8_041details
Qianfan OCR4.7BF1610.1 GBQ8_041details
Qwen3.5 4B4.7BF169.2 GBF1642details
SingVL S1 4B4.5BF169.0 GBF1643details
LightOnOCR 3 4B4.5BF169.0 GBF1643details
Intern Decision 4B4.5BF169.0 GBF1643details
Qwen3 VL 4B Instruct4.4BF169.3 GBQ8_045details
Gemma 3 4B IT4.3BF168.4 GBF1646details
MedGemma 1.5 4B IT4.3BF168.4 GBF1646details
MedGemma 4B IT4.3BF168.4 GBF1646details
TranslateGemma 4B IT4.3BF169.2 GBF1642details
Gemma 3 4B PT4.3BF168.4 GBF1646details
Phi 3.5 Vision Instruct4.1BQ8_07.6 GBno82details
Qwen3 4B4.0BF169.3 GBQ8_045details
Qwen3 4B Instruct 25074.0BF169.3 GBQ8_045details
Granite Vision 4.1 4B4.0BF167.6 GBF1653details
Phi 4 Mini Instruct3.8BF168.9 GBQ8_047details
Phi 3 Mini 4K Instruct3.8BQ8_07.2 GBno89details
Phi 3.5 Mini Instruct3.8BQ8_07.2 GBno89details
Phi 3 Mini 128K Instruct3.8BQ8_07.2 GBno89details
Qwen2.5 VL 3B Instruct3.8BF168.0 GBF1648details
Granite 4.2 3B3.7BF168.0 GBF1649details
Granite 4.1 3B Llmon Preview3.7BF168.0 GBF1649details
Raon 1 Nano3.5BF167.0 GBF1657details
Granite 4.1 3B3.4BF167.6 GBF1653details
Granite 4.0 Micro3.4BF167.6 GBF1653details
DeepSeek OCR 23.4B (1.2B act.)F166.6 GBF16174details
DeepSeek VL2 Tiny3.4B (1.2B act.)F167.4 GBF16153details
Tiny Aya Global3.3BF167.5 GBF1654details
Unlimited OCR 23.3B (1.1B act.)F166.6 GBF16178details
Unlimited OCR3.3B (1.1B act.)F166.6 GBF16179details
DeepSeek OCR3.3B (1.1B act.)F166.6 GBF16179details
Granite 3.1 3B A800M Instruct3.3B (883M act.)F167.1 GBF16204details
Llama 3.2 3B Instruct3.2BF167.5 GBF1656details
Llama 3.2 3B3.2BF167.5 GBF1656details
Hermes 3 Llama 3.2 3B3.2BF167.5 GBF1656details
d1 3B3.1BF165.8 GBF1667details
LFM2.5 VL 3B3.1BF165.8 GBF1667details
Qwen2.5 3B Instruct3.1BF166.7 GBF1658details
SmolLM3 3B3.1BF166.9 GBF1658details
SmolLM3 3B Base3.1BF166.9 GBF1659details
Phi 22.8BF168.2 GBno65details
LFM2.5 2.6B2.7BQ8_03.4 GBQ8_0125details
Granite 3.3 2B Instruct2.5BF165.8 GBF1671details
MiniCPM5 2B2.5BF165.7 GBF1671details
WeVisOne VL 2B2.4BF165.4 GBF1688details
Cosmos Reason2 2B2.4BF165.4 GBF1688details
Qwen3.5 2B2.3BF164.6 GBF1692details
SmolVLM Instruct2.2BF165.4 GBF1699details
Qwen3 1.7B2.0BF165.4 GBF1688details
d1 2B1.9BF164.0 GBF16104details
LFM3 VL 2B1.9BF164.0 GBF16104details
DeepSeek R1 Distill Qwen 1.5B1.8BF164.2 GBF16101details
LFM3 1.7B1.7BF167.2 GBF1652details
LFM3 1.7B Base1.7BF164.0 GBF16104details
SmolLM2 1.7B Instruct1.7BF165.2 GBF16105details
SmolLM2 1.7B1.7BF165.2 GBF16105details
SmolLM 1.7B1.7BF165.2 GBF16105details
Granite 4.0 1B1.6BF164.3 GBF16110details
Qwen2.5 1.5B Instruct1.5BF163.8 GBF16117details
OLMo 2 0425 1B1.5BF164.4 GBF16121details
DeepSeek Coder 1.3B Instruct1.3BF164.5 GBF16134details
Llama 3.2 1B1.2BF163.2 GBF16145details
Llama 3.2 1B Instruct1.2BF163.2 GBF16145details
LFM2.5 1.2B Instruct1.2BF165.0 GBF1677details
LFM2 1.2B1.2BF163.0 GBF16154details
HunyuanOCR1.1BF163.5 GBF16161details
PaddleOCR VL 1.6959MF162.5 GBF16188details
Qwen3.5 0.8B873MF162.4 GBF16231details
LightOnOCR 3 0.8B853MF162.4 GBF16237details
FastVLM 0.5B759MF162.2 GBF16237details
Qwen3 0.6B752MF163.0 GBF16239details
GOT OCR2.0716MF162.8 GBF16251details
Nemotron 3.5 Parse631MF161.9 GBF16275details
GOT OCR 2.0 HF561MF162.5 GBF16321details
SmolVLM2 500M Video Instruct507MF161.6 GBF16439details
SmolVLM 500M Instruct507MF161.6 GBF16439details
Qwen2.5 0.5B Instruct494MF161.7 GBF16364details
LFM2.5 VL 450M449MF161.3 GBF16506details
SmolLM2 360M Instruct362MF161.5 GBF16496details
SmolLM2 360M362MF161.5 GBF16496details
LFM2.5 350M354MF161.9 GBF16253details

Bigger models with CPU offload

These popular models don't fit in the RTX 3060 12GB's VRAM at 4-bit, but run with llama.cpp offload, keeping part of the model in system RAM. Speed range is DDR4-3200 to DDR5-6000 (estimate). Method.

ModelParamsOffload plan
GLM 5.3 Flash321B MoEOffload Q2_K, 128 GB RAM · ~6.7–13 tok/s
gpt-oss 120B117B MoEOffload Q4_K_M, 64 GB RAM · ~13–25 tok/s
DeepSeek V4 Flash 0731304B MoEOffload Q2_K, 128 GB RAM · ~7.0–12 tok/s
DeepSeek V4 Flash291B MoEOffload Q2_K, 128 GB RAM · ~8.9–17 tok/s
GLM 4.5 Air110B MoEOffload Q4_K_M, 128 GB RAM · ~5.0–9.3 tok/s
MiniMax M2.5229B MoEOffload Q3_K_M, 128 GB RAM · ~6.6–13 tok/s
Hy3299B MoEOffload Q3_K_M, 128 GB RAM · ~4.0–7.7 tok/s
Nex N2.6321B MoEOffload Q2_K, 128 GB RAM · ~6.4–12 tok/s

Comparing cards? The RTX 3060 12GB (launch MSRP $329) is in best GPUs for local LLMs under $500. See also Mac vs PC.

Just too big at 4-bit

These need up to 35% more memory than the RTX 3060 12GB has at Q4_K_M and 8K context. A smaller quant or shorter context may work.

ModelQ4_K_M needsShort by
gpt-oss Safeguard 20B11.8 GB0.4 GB
gpt-oss 20B11.8 GB0.4 GB
DeepSeek MoE 16B Base12.5 GB1.1 GB
ERNIE 4.5 21B A3B PT13.6 GB2.2 GB
ERNIE 4.5 21B A3B Thinking13.6 GB2.2 GB
LFM2.5 24B A2B14.3 GB2.9 GB

Questions

What is the biggest model that runs on the RTX 3060 12GB?

Of the models we track, the largest that fits at 8K context is GLM 4.7 Flash (31B) at Q2_K, needing about 11.4 GB.

How much of the RTX 3060 12GB's memory can a model use?

About 11.4 GB: we assume 95% of the 12 GB is usable after driver and display overhead.

Spec source: NVIDIA specifications. Model data: Hugging Face Hub API. Speed is a bandwidth ceiling (bandwidth ÷ bytes read per token), not a benchmark. Methodology · Report an error.