SkywardAI is an open-source community dedicated to developing efficient and specialized AI models. Founded by passionate students from RMIT University Melbourne, our mission is to create high-performing AI solutions that prioritize accessibility, efficiency, and sustainability. We empower users to run advanced AI models on consumer-grade hardware while supporting cloud-native deployment.
The community is inspired by an open-source OpenAI alternative API LocalAI.
Our AI open-source community aims to develop specialized, high-performing models that are not only smaller but also optimized for specific tasks. Unlike massive LLMs, these models prioritize efficiency and accessibility while maintaining top-tier performance. By minimizing data requirements and reducing computational demands, we can create a more sustainable AI ecosystem—maximizing intelligence per watt. This approach ensures that AI remains both powerful and practical, making advanced technology more accessible for widespread adoption.
Our mission is to make AI both powerful and practical by developing efficient, accessible, and sustainable models.
We aim to deliver:
- Highly efficient, specialized AI models optimized for specific tasks
- Smooth streaming inference on CPUs with open-source multimodal models under 10B parameters
- Inference with RAG mode for enhanced retrieval-augmented generation
- A simple trainer for customizing neural net training
- One-command deployment for seamless integration
Our vision is to create a thriving community that drives the future of AI with sustainability and efficiency at its core.
- AI should be both accessible and efficient—maximizing intelligence per watt
- Everyone can be a neural network hero—today we define functions, tomorrow we define neural networks
- AI isn't just a trend; it should be a practical, deployable solution on consumer-grade hardware
- By reducing data requirements and computational overhead, we ensure AI remains sustainable and widely available
