Heat is a native iOS and macOS client designed to provide a unified interface for interacting with various Large Language Models (LLMs), both cloud-hosted and locally run. It aims to simplify access to popular LLM services and open-source models for users who want to leverage AI capabilities across their Apple devices.

How It Works

Heat utilizes a companion library, Swift GenKit, to abstract the complexities and differences between various LLM providers, including OpenAI, Mistral, Perplexity, Anthropic, and local Ollama deployments. This abstraction allows Heat to offer a consistent user experience regardless of the underlying model or service, supporting features like multi-step tool use, web search integration, and calendar/filesystem access.

Quick Start & Requirements

• Install: Build and run directly in Xcode.
• Prerequisites: API keys for cloud services (OpenAI, Mistral, etc.). For local Ollama, install Ollama and pull models. iOS app requires local IP configuration for Ollama server access.
• Resources: Requires an Apple Developer account for building on physical devices.

Highlighted Details

• Supports major LLM providers (OpenAI, Mistral, Anthropic, Gemini) and local Ollama.
• Includes image generation capabilities (Stable Diffusion, DALL-E).
• Features a Spotlight-like launcher (Option + Space).
• Integrates web search, calendar, and filesystem access for enhanced responses.
• No server dependencies beyond the LLM models themselves.

Maintenance & Community

The project is maintained by nathanborror. Further community engagement channels are not explicitly detailed in the README.

Licensing & Compatibility

The README does not specify a license. Compatibility for commercial use or closed-source linking is not detailed.

Limitations & Caveats

The project is currently limited to Apple's ecosystem (iOS and macOS). The original goal of on-device model execution has been deferred due to technical challenges.