HOSTKEY Company blog

Configuring a custom model with tools and RAG in OpenWebUI is easy. Making it behave predictably in production is the hard part. Our support bot powered by Gemma 4 initially provided plausible answers, but they lacked actual system data. We analyzed why this happens and transformed a "single model in a UI" into a deterministic pipeline using Pipes.

We compared LiteLLM and OmniRoute under load: latency, requests per second, memory and CPU usage, and the settings worth changing right after installation.

The same RTX 4090 with a €150 price gap: what are you really paying for? Learn how to tell a whole GPU from a slice of one, when a GPU VPS is enough, and when it is time to move to a dedicated GPU server.

Step-by-step guide to running DeepSeek, Gemma, and Phi on your own server. Ollama, vLLM, Docker, GPU configuration, and code examples.

The debate over how to install software on a server usually ends in a stalemate because everyone has their own habits. We categorized twelve methods, ran five of them on identical servers with a stopwatch, and discovered that more than half of the time difference is not spent installing the application itself, but installing the tool itself.