Caleta LLM Appliance

The Caleta LLM Appliance is a virtual machine image you deploy into your own Microsoft Azure subscription to run private AI. All inference runs on your VM, and your prompts, chat history, outputs and documents never leave your Azure environment.

The appliance is available through the Microsoft Azure Marketplace. Models are downloaded by you, directly from Hugging Face, under each model's own licence; no model weights ship inside the image. The appliance binds to localhost by default, requires no inbound access to operate, and can run fully air-gapped once its models are in place.

Trust and legal

• Privacy statement – what stays in your subscription and the connections the appliance makes
• Terms of service – billing, model licensing, warranties and support
• Security – security posture and vulnerability disclosure

For the wider private AI service, see Private AI in your own Azure tenant, or get in touch.