lvladikov / text-to-image updated 5 days ago

Krea2-Turbo-Distill-4step-LoRA

Name: Krea2-Turbo-Distill-4step-LoRA
Author: lvladikov

Half the steps · ~1.6× faster · 45% of the gap to the 8-step teacher closed · texture at 1.03× the teacher's, verified clean · teacher preferred on only 6 of 45 judged renders · 12 trained resolutions · 78,000 training samples · 21 days on one RTX 3090.

Params

—

Context

—

Downloads 30d

32 K

Likes

170

Licence: other Not gated SAFETENSORS View on Hugging Face ↗

Download history

daily snapshots · 13 days

32 K19 K

Sep 8Sep 12Sep 16Sep 20

Can you run it?

Estimated VRAM at 8K context unless noted. Pick your hardware to see the verdict per quantization.

File	Quant	Size	Est. VRAM	Verdict on RTX 4090 · 24 GB
model.safetensors	fp16	9.6 GB	11.1 GB	✅ Runs comfortably

Estimate: file size × 1.1 + KV cache at 8K + 0.5 GB overhead. Not a benchmark — how we calculate this.

Specifications

Licence: other
First seen on the Hub: 2026-08-20
Base model: Krea-2-Turbo
Training datasets: undisclosed
Added to our catalog: 2026-09-08

Family

Base model and the most-downloaded derivatives in the catalog.

Base Krea-2-Turbo krea 74 K

Compare with any text-to-image model

Popular comparisons

stable-diffusion-xl-base-1.0 3.0 M stable-diffusion-v1-5 1.8 M dreamshaper-7 901 K Juggernaut-XL-v9 868 K sdxl-turbo 853 K FLUX.1-dev 765 K