MDST Engine: run GGUF models in the browser with WebGPU/WASM (mdst.app)

MDST has announced a groundbreaking engine that enables users to run GGUF models directly in their browsers using WebGPU and WebAssembly (WASM). This development is significant as it empowers individuals to create, edit, and collaborate on AI tasks locally without relying on cloud services or complex installations. By making local inference accessible through a simple browser interface, MDST caters to the growing demand for privacy, ease of use, and autonomy in AI model usage, aligning with the trend towards more local computing solutions. The MDST engine offers an integrated, agentic IDE that supports real-time project synchronization and end-to-end encryption, ensuring user privacy and compliance with GDPR. Users can benchmark their models locally against a public leaderboard, which ranks performance based on task difficulty, thus facilitating meaningful comparisons in AI research and development. With support for various cloud and local LLMs in a user-friendly format, the platform promises to enhance the AI/ML community's collaborative capabilities and encourage experimentation using Qwen, Ministral, and other popular models. As mainstream hardware continues to evolve, MDST positions itself at the forefront of making AI more accessible and practical for everyday users.