Zeraix

Zeraix is a local AI desktop for people who want models to work with their files and carry out multi-step tasks. It combines chat, project tools, and agents in one workspace.

• Local models: Run supported GGUF models through llama.cpp-based runtimes, with hardware recommendations and memory estimates; choices include Qwen3.6, Gemma 4, and LFM2.5.
• Project work: Ask agents to read, search, and edit files, review diffs, run terminal commands, and use browser tools within a conversation.
• Agent controls: Delegate tasks to specialized sub-agents, inspect or stop their work, and choose approval controls, including a Plan mode that blocks file changes.
• Continuing tasks: Keep conversations and memory locally, track goals, and schedule recurring workflows with run history and notifications.
• Documents and connections: Use bundled tools for Word documents, PDFs, presentations, and spreadsheets, or extend the workspace through plugins, local or remote MCP servers, and OpenAI-compatible endpoints.

Local inference and local tools work offline after the initial model and runtime downloads. The local core needs no account, subscription, or API key. Local models process data on your device, and conversations stay there. Cloud models, web search, remote tools, and image or video generation services need network access and send the data required for each action to the selected provider.

Zeraix is open source under Apache 2.0 and runs on macOS 13 or later with Apple Silicon, and Windows 10/11 x64. It recommends at least 16 GB of memory, though some smaller models support 8 GB systems. Model capabilities and context length depend on the runtime, platform, and available memory. An optional QEMU sandbox isolates supported commands; host commands run directly on your computer.