Osaurus
Native macOS AI harness enabling local and cloud model execution with persistent memory, autonomous agents, and cryptographic identity—fully offline and open source.
Community:
Product Overview
What is Osaurus?
Osaurus is a native macOS application built in Swift that serves as a local-first AI runtime and agent harness. It sits between users and any AI model—whether running locally on Apple Silicon or connected to cloud providers—and provides the continuity layer that makes AI personal: persistent memory, autonomous task execution, and cryptographic identity. The platform supports Apple's on-device Foundation models, local MLX models, and cloud providers including OpenAI, Anthropic, xAI, and OpenRouter, all while keeping files, tools, and memory on the user's hardware.
Key Features
Local-First Model Execution
Run AI models entirely on your Mac using Apple Silicon, MLX, or Ollama—no internet required. Supports Apple Foundation models, Gemma, Qwen, Llama, DeepSeek, and Liquid AI's LFM family for complete offline operation.
Persistent Memory & Identity
Cryptographic identity stored in iCloud Keychain with Face ID/Touch ID protection, plus 24-word recovery phrase. Distills conversations into compact facts for long-term recall across all sessions and agents.
Autonomous Agent Framework
Build agents that execute tasks, use tools, and work in parallel. Agents can read/write files, run scripts in isolated Linux VMs (Sandbox), and operate via voice control or folder watchers.
Unified Cloud & Local Integration
Seamlessly switch between local models and cloud providers (OpenAI, Anthropic, Gemini, xAI, OpenRouter, LM Studio) within one interface, carrying shared memory across all models.
Open Source & Developer Tools
MIT licensed with full source access. Includes HTTP API (OpenAI/Anthropic/Ollama compatible), SDK examples in Python and JavaScript, CLI commands, and extensible tools/plugins system.
Privacy-First Architecture
No telemetry, no data training, no servers in the loop. All voice transcription, model inference, and file operations happen on-device with zero data leaving your Mac unless explicitly configured.
Use Cases
- Private Code Development : Developers can run AI-assisted coding, debugging, and documentation entirely offline, keeping proprietary code and development contexts completely private on their local machine.
- Autonomous Task Automation : Build agents that monitor folders, execute scripts, summarize files, and generate artifacts like READMEs while you sleep—running in isolated VMs for security.
- Personal Knowledge Management : Create specialized agents with persistent memory for different domains (writing, research, coding) that remember your preferences, past conversations, and work context across sessions.
- Voice-Controlled Workflows : Use on-device voice transcription and wake-words to dictate into any app, control agents via voice commands, or set up global transcription mode for hands-free operation.
- Hybrid Model Workflows : Start with fast local models for routine tasks, then seamlessly escalate to frontier cloud models for complex reasoning—maintaining conversation continuity and memory across both.
FAQs
Osaurus Alternatives
Applied Compute
Cloud platform for enterprises to post-train, serve, and continuously improve custom open foundation models .
Zyphra
AI company developing advanced multimodal agent systems and high-quality datasets to power efficient, small-scale language models.
Bluesminds
Enterprise AI intelligence platform with unified API access to 200+ LLM models, autonomous agents, and sovereign deployment across 29 global regions.
Google DeepMind
Leading AI research lab pioneering general-purpose AI technologies to solve complex scientific and real-world problems.
硅基流动
Comprehensive cloud platform providing high-performance inference services for large language models and image generation with cost-effective APIs.
LongCat API Platform
API platform providing access to LongCat series models with OpenAI and Anthropic compatibility, featuring 1M context window and high-throughput agentic capabilities.
Crusoe Cloud
Energy-efficient AI cloud infrastructure platform combining renewable-powered data centers with optimized GPU compute and managed inference services for accelerated model deployment.
ATXP
Infrastructure protocol that gives AI agents a persistent account with identity, payments, email, and access to 14+ tools — all pay-as-you-go, no subscriptions needed.