Local AI
A local AI workbench for open-model work on my own hardware.
I use it to run, compare, and learn from open models across text, vision, image generation, embeddings, and chat.
Local workflows supported today.
Text Models
Chat, coding, summarization, and general language work.
Vision Models
Inspect screenshots, diagrams, images, and visual documents.
Image Generation
Create and edit images with local diffusion workflows.
Embedding Models
Build vectors for search, retrieval, and knowledge experiments.
Chat UI
Ask, compare, and review model outputs in one private interface.
A controlled space for hands-on learning.
I use frontier cloud models heavily. Local AI gives me a private space for open-model exploration.
Useful for:
- testing model behavior across text, vision, images, and embeddings
- building prototypes without sending every run to an API
- learning how inference, retrieval, and orchestration work together
Models organized around the work they support.
Text, vision, image, and embedding tools stay easy to find, so experiments start with the right kind of model.
Model names and logos are used descriptively to identify compatible model families. No sponsorship or endorsement implied.
Compare responses without changing tools.
I can switch models inside the same conversation and see how each one handles a prompt.
Estimated API savings since April 2026
After the hardware investment, repeated tests, image runs, embedding jobs, and prototypes do not create per-token or per-image API charges.
The estimate is directional, not accounting. It shows how much experimentation local compute has enabled.