Local AI Image Tools
Use your own machine's AI to understand, generate and edit images. Connect to your local Ollama or Stable Diffusion — nothing leaves your device.
All requests go directly from your browser to your local Ollama or Stable Diffusion server. No image, prompt or response is ever sent to our servers.
Vision (Ollama)
Generate (Stable Diffusion)
Quick Start
Install Ollama
Download from ollama.com and install. macOS / Linux: brew install ollama or use the one-line installer.
Pull a vision model
Run: ollama pull llava (or qwen2.5vl, llama3.2-vision). Needs 5-8 GB disk space.
Start with CORS enabled
Run: OLLAMA_ORIGINS="*" ollama serve — then click "Test connection" above.
Ollama settings
Upload an image
Click or drag an image here
What to do with this image
Custom prompt
Result
Upload an image and run a task to see the result here.
What are local AI image tools?
Local AI image tools run open-source vision and image-generation models on your own computer. This page connects directly to those local servers from your browser, so prompts and images never reach our servers.
Cloud AI vs Local AI
| Aspect | Cloud AI services | Local AI (this page) |
|---|---|---|
| Privacy | Images uploaded to a vendor server | Images stay on your device |
| Cost | Per-request fees or monthly subscription | Free after one-time install |
| Speed | Bound by network and vendor queue | Bound by your CPU/GPU; typically 5–60s per image |
| Model choice | Limited to vendor offerings | Any open-source model you can run |
Supported model categories
Vision models (image understanding)
Models like LLaVA, Qwen2-VL or Llama 3.2 Vision describe images, extract text (OCR), suggest tags or alt text. Typical disk size: 4.7 GB (LLaVA 7B) to 8 GB+ (LLaVA 13B). Pull them with Ollama.
Image-generation models
Stable Diffusion 1.5, SDXL and Forge variants generate new images from text prompts and edit existing ones (img2img, inpainting). Typical checkpoint size: 2–4 GB (SD 1.5) to 6–7 GB (SDXL).
Quickstart in 3 steps
- Install Ollama (for vision) or Stable Diffusion WebUI (for generation) on your computer.
- Pull a model and start the local server with CORS enabled for this site.
- Open the Vision or Generate tab below and click Test connection.