Local LLMs

Host large language models directly within your DaDesktop environment rather than routing prompts to external API services. This setup leverages DaDesktop's dedicated GPU resources to process requests, ensuring that your prompts, files, and sensitive data remain strictly confined to your private workspace. In this model, you pay for the allocated desktop infrastructure rather than incurring variable costs per token generated.

Why run LLMs locally

• Enhanced Data Privacy: All prompts, documents, and associated data are securely contained within the DaDesktop environment, mitigating the risks associated with third-party data transmission.
• Predictable Cost Structure: Utilize DaDesktop's high-performance GPU infrastructure without the financial burden of per-request API fees, allowing for more budget-friendly scaling.
• Model Flexibility: Select specific models that best suit your requirements and maintain full control over their configuration and deployment settings.
• Seamless Workload Integration: Deploy models natively alongside your other applications and development tools, enabling a cohesive workflow without external dependencies.

How it works

To get started, select a DaDesktop instance equipped with GPU capabilities, load your chosen model, and initiate local execution. Thanks to GPU passthrough technology, your desktop gains direct access to physical GPU hardware, optimizing performance for LLM-intensive tasks. You can review available GPU options, technical specifications, and supported configurations on the GPU page.