What can NVIDIA GeForce RTX 3060 12GB run locally?

26 models fit comfortably on NVIDIA GeForce RTX 3060 12GB with 32 GB system RAM. 166 more run with tradeoffs. Top pick: Ministral 3B Instruct.

What's the recommended runner for this hardware?

llama.cpp / Ollama (CUDA). This is the runner that gives best tokens-per-second on NVIDIA GeForce RTX 3060 12GB based on windows support and the GPU backend.

What can NVIDIA GeForce RTX 3060 12GB run for long context?

Build: RTX 3060 12GB + Ryzen 5 5600 + 32GB DDR4 (cheapest path)

Memory: 12 GB VRAM + 32 GB system RAM

Runner: llama.cpp / Ollama (CUDA)

Any Chat Coding Agents Reasoning Vision Long context Creative

Runs comfortably
26 models

Ranked by fit for long context use case + predicted speed. Click a row for VRAM breakdown.

#1Ministral 3B Instruct

mistral

Quant: Q4_K_MContext: 8,192VRAM: 5.3 GBHeadroom: 6.7 GBTTFT: noticeable

129

tok/s

Estimated

Weights

1.90 GB

KV cache

1.50 GB

Activations

0.10 GB

Runtime

1.80 GB

Time to first token (prefill, 512-token prompt): ~605 ms (noticeable)

Model details →

#2Qwen 2.5 7B Instruct

qwen

Commercial OK

Quant: Q5_K_MContext: 8,192VRAM: 7.9 GBHeadroom: 4.1 GBTTFT: noticeable

ollama run qwen2.5:7b

tok/s

Estimated

Weights

5.40 GB

KV cache

0.47 GB

Activations

0.28 GB

Runtime

1.80 GB

Time to first token (prefill, 512-token prompt): ~1411 ms (noticeable)

Model details →

#3Hermes 3 Llama 3.2 3B

hermes

Commercial OK

Quant: Q4_K_MContext: 8,192VRAM: 5.2 GBHeadroom: 6.8 GBTTFT: noticeable

129

tok/s

Estimated

Weights

1.80 GB

KV cache

1.50 GB

Activations

0.10 GB

Runtime

1.80 GB

Time to first token (prefill, 512-token prompt): ~605 ms (noticeable)

Model details →

#4Dolphin 3.0 Llama 3.2 3B

dolphin

Commercial OK

Quant: Q4_K_MContext: 8,192VRAM: 5.2 GBHeadroom: 6.8 GBTTFT: noticeable

129

tok/s

Estimated

Weights

1.80 GB

KV cache

1.50 GB

Activations

0.10 GB

Runtime

1.80 GB

Time to first token (prefill, 512-token prompt): ~605 ms (noticeable)

Model details →

#5Phi-4 Mini 4B

3.8B

phi

Commercial OK

Quant: Q4_K_MContext: 8,192VRAM: 6.2 GBHeadroom: 5.8 GBTTFT: noticeable

102

tok/s

Estimated

Weights

2.40 GB

KV cache

1.90 GB

Activations

0.13 GB

Runtime

1.80 GB

Time to first token (prefill, 512-token prompt): ~766 ms (noticeable)

Model details →

#6Phi-4 Reasoning Mini 4B

3.8B

phi

Commercial OK

Quant: Q4_K_MContext: 8,192VRAM: 6.2 GBHeadroom: 5.8 GBTTFT: noticeable

102

tok/s

Estimated

Weights

2.40 GB

KV cache

1.90 GB

Activations

0.13 GB

Runtime

1.80 GB

Time to first token (prefill, 512-token prompt): ~766 ms (noticeable)

Model details →

#7Phi-3.5 Vision

4.2B

phi

Commercial OK

Quant: Q4_K_MContext: 8,192VRAM: 6.5 GBHeadroom: 5.5 GBTTFT: noticeable

tok/s

Estimated

Weights

2.50 GB

KV cache

2.10 GB

Activations

0.13 GB

Runtime

1.80 GB

Time to first token (prefill, 512-token prompt): ~847 ms (noticeable)

Model details →

#8Llama 3.2 3B Instruct

llama

Commercial OK

Quant: Q8_0Context: 8,192VRAM: 6.9 GBHeadroom: 5.1 GBTTFT: noticeable

ollama run llama3.2:3b

tok/s

Estimated

Weights

3.40 GB

KV cache

1.50 GB

Activations

0.18 GB

Runtime

1.80 GB

Time to first token (prefill, 512-token prompt): ~605 ms (noticeable)

Model details →

#9Falcon 3 3B Instruct

falcon

Commercial OK

Quant: Q4_K_MContext: 8,192VRAM: 5.1 GBHeadroom: 6.9 GBTTFT: noticeable

129

tok/s

Estimated

Weights

1.70 GB

KV cache

1.50 GB

Activations

0.09 GB

Runtime

1.80 GB

Time to first token (prefill, 512-token prompt): ~605 ms (noticeable)

Model details →

#10OpenELM 3B Instruct

other

Quant: Q4_K_MContext: 2,048VRAM: 4.0 GBHeadroom: 8.0 GBTTFT: noticeable

129

tok/s

Estimated

Weights

1.70 GB

KV cache

0.38 GB

Activations

0.09 GB

Runtime

1.80 GB

Time to first token (prefill, 512-token prompt): ~605 ms (noticeable)

Model details →

#11StarCoder 2 3B

other

Commercial OK

Quant: Q4_K_MContext: 8,192VRAM: 5.4 GBHeadroom: 6.6 GBTTFT: noticeable

129

tok/s

Estimated

Weights

2.00 GB

KV cache

1.50 GB

Activations

0.11 GB

Runtime

1.80 GB

Time to first token (prefill, 512-token prompt): ~605 ms (noticeable)

Model details →

#12SmolLM 3 3B

other

Commercial OK

Quant: Q4_K_MContext: 8,192VRAM: 5.2 GBHeadroom: 6.8 GBTTFT: noticeable

129

tok/s

Estimated

Weights

1.80 GB

KV cache

1.50 GB

Activations

0.10 GB

Runtime

1.80 GB

Time to first token (prefill, 512-token prompt): ~605 ms (noticeable)

Model details →

Runs with tradeoffs
166 models

Tight VRAM, partial CPU offload, or context-limited.

Phi-3.5 Mini Instruct

3.8B

phi

Commercial OK

Quant: Q8_0Context: 8,192VRAM: 8.0 GBHeadroom: 4.0 GBTTFT: noticeable

• Tight VRAM fit — only 4.0 GB headroom left for context growth

ollama run phi3.5:3.8b

tok/s

Estimated

Weights

4.10 GB

KV cache

1.90 GB

Activations

0.21 GB

Runtime

1.80 GB

Time to first token (prefill, 512-token prompt): ~766 ms (noticeable)

Model details →

Falcon Mamba 7B

falcon

Commercial OK

Quant: Q4_K_MContext: 8,192VRAM: 9.7 GBHeadroom: 2.3 GBTTFT: noticeable

• Tight VRAM fit — only 2.3 GB headroom left for context growth

tok/s

Estimated

Weights

4.20 GB

KV cache

3.50 GB

Activations

0.22 GB

Runtime

1.80 GB

Time to first token (prefill, 512-token prompt): ~1411 ms (noticeable)

Model details →

Codestral Mamba 7B

mistral

Commercial OK

Quant: Q4_K_MContext: 8,192VRAM: 9.7 GBHeadroom: 2.3 GBTTFT: noticeable

• Tight VRAM fit — only 2.3 GB headroom left for context growth

tok/s

Estimated

Weights

4.20 GB

KV cache

3.50 GB

Activations

0.22 GB

Runtime

1.80 GB

Time to first token (prefill, 512-token prompt): ~1411 ms (noticeable)

Model details →

Gemma 4 E4B (Effective 4B)

gemma

Commercial OK

Quant: Q8_0Context: 8,192VRAM: 8.4 GBHeadroom: 3.6 GBTTFT: noticeable

• Tight VRAM fit — only 3.6 GB headroom left for context growth

ollama run gemma4:e4b

tok/s

Estimated

Weights

4.40 GB

KV cache

2.00 GB

Activations

0.23 GB

Runtime

1.80 GB

Time to first token (prefill, 512-token prompt): ~806 ms (noticeable)

Model details →

Qwen 3 8B

qwen

Commercial OK

Quant: Q4_K_MContext: 8,192VRAM: 10.8 GBHeadroom: 1.2 GBTTFT: noticeable

• Tight VRAM fit — only 1.2 GB headroom left for context growth

ollama run qwen3:8b

tok/s

Estimated

Weights

4.80 GB

KV cache

4.00 GB

Activations

0.25 GB

Runtime

1.80 GB

Time to first token (prefill, 512-token prompt): ~1613 ms (noticeable)

Model details →

Command R7B (12-2024)

command-r

Quant: Q4_K_MContext: 8,192VRAM: 10.4 GBHeadroom: 1.6 GBTTFT: noticeable

• Tight VRAM fit — only 1.6 GB headroom left for context growth

tok/s

Estimated

Weights

4.40 GB

KV cache

4.00 GB

Activations

0.23 GB

Runtime

1.80 GB

Time to first token (prefill, 512-token prompt): ~1613 ms (noticeable)

Model details →

Ministral 8B Instruct

mistral

Quant: Q4_K_MContext: 8,192VRAM: 11.1 GBHeadroom: 0.9 GBTTFT: noticeable

• Tight VRAM fit — only 0.9 GB headroom left for context growth

tok/s

Estimated

Weights

5.00 GB

KV cache

4.00 GB

Activations

0.26 GB

Runtime

1.80 GB

Time to first token (prefill, 512-token prompt): ~1613 ms (noticeable)

Model details →

Llama 3.1 8B Instruct

llama

Commercial OK

Quant: Q5_K_MContext: 8,192VRAM: 8.9 GBHeadroom: 3.1 GBTTFT: noticeable

• Tight VRAM fit — only 3.1 GB headroom left for context growth

ollama run llama3.1:8b

tok/s

Estimated

Weights

5.70 GB

KV cache

1.07 GB

Activations

0.29 GB

Runtime

1.80 GB

Time to first token (prefill, 512-token prompt): ~1613 ms (noticeable)

Model details →

What if you upgraded?

Hypothetical scenarios. We re-ran the compatibility engine for each.

+32 GB system RAM

~$80–150

Doubles your CPU-offload working set. Helps when models don't quite fit in VRAM.

Unlocks: 41 new comfortable, 174 new tradeoff

• Qwen 3 0.6B
• Qwen 3 1.7B
• Gemma 3 270M
• SmolLM2 135M Instruct

Shop this upgrade↗

Upgrade to NVIDIA GeForce RTX 4070 Ti Super

~$829

16 GB VRAM (vs your 12 GB) plus a bandwidth jump from ~360 GB/s to ~672 GB/s.

Unlocks: 116 new comfortable

• Qwen 3 0.6B
• Llama 3.1 8B Instruct
• Qwen 3 8B
• Qwen 3 1.7B

Shop this upgrade↗

Add a second NVIDIA GeForce RTX 3060 12GB

~$249

Tensor parallelism splits the model across both cards, effectively doubling VRAM. Bandwidth doesn't double — runs ~1.5× the single-card speed in practice.

Unlocks: 158 new comfortable

• Qwen 3 0.6B
• Llama 3.1 8B Instruct
• Qwen 3 8B
• Gemma 4 12B

Shop this upgrade↗

Some links above are affiliate links. We may earn a commission at no extra cost to you. How we make money.

Won't run
top 5 popular models

Need more memory than you have. Shown for orientation.

DeepSeek V4 Pro (1.6T MoE)

1600B

deepseek

Commercial OK

Even with CPU offload, needs more memory than your VRAM (12 GB) + 60% of system RAM (19 GB) combined.

—

Qwen 3.5 235B-A17B (MoE)

397B

qwen

Commercial OK

Even with CPU offload, needs more memory than your VRAM (12 GB) + 60% of system RAM (19 GB) combined.

—

Qwen 3 235B-A22B

235B

qwen

Commercial OK

Even with CPU offload, needs more memory than your VRAM (12 GB) + 60% of system RAM (19 GB) combined.

—

DeepSeek R1 (671B reasoning)

671B

deepseek

Commercial OK

Even with CPU offload, needs more memory than your VRAM (12 GB) + 60% of system RAM (19 GB) combined.

—

Llama 4 Scout

109B

llama

Commercial OK

Even with CPU offload, needs more memory than your VRAM (12 GB) + 60% of system RAM (19 GB) combined.

—

How to read these numbers

Measured here

Measured here - RunLocalAI ran this exact combo on owner hardware with public evidence.

Source-backed

Source-backed / community - a reproduced public source supports the speed, but it is not labeled as owner-measured.

Extrapolated

Extrapolated - predicted from a measured benchmark on similar-bandwidth hardware.

Estimated

Estimated - formula based on VRAM bandwidth and model architecture; not a benchmark row.

RunLocalAI Will-It-Run Framework →

Want a specific benchmark we don't have? Email Contact support and we'll prioritize it.

Runs comfortably26 models

Runs with tradeoffs166 models

What if you upgraded?

+32 GB system RAM

Upgrade to NVIDIA GeForce RTX 4070 Ti Super

Add a second NVIDIA GeForce RTX 3060 12GB

Won't runtop 5 popular models

How to read these numbers

Runs comfortably
26 models

Runs with tradeoffs
166 models

Won't run
top 5 popular models