Is 12GB VRAM enough for local LLMs?
Yes for models up to about 15B at Q4_K_M with 8k context (est.); the largest dense fit is Phi 4 reasoning vision 15B. 18 of 64 current models fit at Q3_K_M or better. 27B to 32B dense models and 70B do not fit without RAM offload.
Checked 2026-10-22 · memory figures are estimates
| Model | Needs (Q4_K_M, 8k, est.) | |
|---|---|---|
| Largest dense | Phi 4 reasoning vision 15B | 11.9 GB |
| Largest overall | Phi 4 reasoning vision 15B | 11.9 GB |
How we got this number
Memory needed = parameters × bits per weight ÷ 8 (Q4_K_M is 4.85 bits, Q8_0 is 8.5), plus the KV cache for your context and runtime buffers. Worked example: Phi 4 reasoning vision 15B has 15.12B parameters, so Q4_K_M weights are about 9.2 GB; with an 8k f16 KV cache (1.7 GB) and buffers the total is about 11.9 GB. All estimates. Full method.
FAQ
Is 24GB VRAM enough for AI?
For running models, 24 GB fits 27B to 32B dense models at Q4 with short context (est.). See the 24 GB list.
Can 12 GB run a 30B MoE?
Only with expert offload to system RAM. llama.cpp can keep attention on the GPU and send routed experts to RAM, at a speed cost.
Which 12 GB cards are common?
RTX 3060 12GB and RTX 5070 are in our list. Check either in the checker with your RAM.
Open the checker · What fits my GPU?
More short answers
- What LLM can I run on my GPU?
- How much VRAM do I need for a 70B model?
- Is 16GB RAM enough to run an LLM locally?
- MLX vs GGUF on a Mac: which is faster?
- How much VRAM do I need for a 7B model?
- How much VRAM do I need for a 32B model?
- How much VRAM does context length use?
- Can you run an LLM on CPU only?
- Q4 vs Q8: which quantization should I use?
Will It Run? Pro NOT BUILT YET
A personal setup for your exact hardware (Ollama / llama.cpp configs ready for your coding agent), an alert when a new model fits, and a one-click cloud fallback when it doesn't.
Not built yet. Join the waitlist, $10/mo planned, no charge now. What's planned
Email only. No payment info.