Google Colab notebooks · free GPU, one click

Curated, runnable notebooks — image generation, fine-tuning, audio, transcription, upscaling. Click Open in Colab, sign in with a Google account, and run the cells. Free Colab gives a GPU for a while; each note says what it needs. 9 notebooks.

image

ComfyUI in Colab image

Run a full ComfyUI server inside Colab with a public tunnel — then load any of our ComfyUI workflow templates. The bridge between the two layers on this page.
GPU: free T4 is enough for SD1.5/SDXL; FLUX wants more

▶ Open in Colab ↗ comfyanonymous/ComfyUI

SDXL with 🤗 Diffusers image

Generate images with SDXL using the Hugging Face Diffusers library — a few cells, very approachable. Good first notebook for text-to-image.
GPU: free T4

▶ Open in Colab ↗ huggingface/notebooks

FLUX.1 with 🤗 Diffusers image

Run FLUX.1 — top-tier prompt following and text-in-image — via Diffusers. Heavier; use the [schnell] variant or quantization on the free tier.
GPU: free T4 only with the schnell/quantized variant; [dev] wants more

▶ Open in Colab ↗ huggingface/notebooks

finetune

DreamBooth LoRA — Fine-tune SDXL on your subject finetune

Train a small LoRA so the model learns a specific person, object, or style from a handful of images. Outputs a tiny .safetensors you load anywhere.
GPU: free T4 works for LoRA; full DreamBooth wants more VRAM

▶ Open in Colab ↗ huggingface/notebooks

audio

MusicGen — Text to Music audio

Generate short music clips from a text description with Meta's MusicGen. Great for blog/reel background cues — pair the output with the reel template maker.
GPU: free T4

▶ Open in Colab ↗ facebookresearch/audiocraft

Bark — Text to Speech / Voice audio

Generate spoken audio (and sound effects) from text with Bark. Useful for voiceover on reels and tutorials.
GPU: free T4

▶ Open in Colab ↗ suno-ai/bark

video

AnimateDiff — Text to short video video

Generate short animated clips from a prompt with AnimateDiff motion modules on top of SD. A free-tier on-ramp to text-to-video.
GPU: free T4 (short clips); more VRAM for longer/HD

▶ Open in Colab ↗ huggingface/notebooks

upscale

Real-ESRGAN — Upscale & restore images upscale

Upscale and clean up photos and art (including a face-enhance mode) with Real-ESRGAN. Fast, no diffusion — a good finishing step.
GPU: free T4 (or CPU, slower)

▶ Open in Colab ↗ xinntao/Real-ESRGAN

text

Whisper — Transcribe audio to text text

Transcribe (and translate) audio with OpenAI Whisper — turn a recording into captions for your reel or a transcript for a post.
GPU: free T4 (or CPU for the small models)

▶ Open in Colab ↗ openai/whisper