Can you train a Krea 2 LoRA on a 12 GB GPU? Here is what the sources actually say.
Not published by trainers. musubi: 12GB+ recommended for image training (general guidance); fp8 or --convrot_int8 halves weight VRAM
My own Krea 2 recipe uses qfloat8 model + text encoder, low_vram, gradient checkpointing and cached latents; see the full recipe. I train it on a rented RTX 5090 (32 GB), so I have not verified it at 12 GB.
Other tiers: Krea 2 overview
Cheapest 12 GB+ cards (2026-10-16 14:43 UTC) from RunPod and Vast.ai. Prices move constantly; see the live price table.
| GPU | VRAM | RunPod community $/hr | Vast.ai cheapest $/hr |
|---|---|---|---|
| RTX 3060 | 12 GB | — | $0.06 |
| RTX A4000 | 16 GB | $0.17 | $0.09 |
| RTX 4070S | 12 GB | — | $0.10 |
| Tesla V100 | 16 GB | $0.19 | $0.10 |
| RTX 3090 | 24 GB | $0.22 | $0.12 |
| RTX 5060 Ti | 16 GB | — | $0.12 |
No trainer publishes a Krea 2-specific number. musubi-tuner's general guidance is 12 GB+ for image training, and its fp8 / ConvRot int8 options halve weight VRAM. Whether 12 GB is comfortable for Krea 2 is unknown.
See the table above and the live price page.
Data updated 2026-10-16. Disclosure: RunPod and Vast.ai links are referral links; ImageLoRA may earn a commission at no cost to you. unknown = not yet verified from a source.