Running DeepSeek on a Worldstream dedicated server
Kort antwoord
Worldstream's DeepSeek dedicated server ships with the 14B DeepSeek R1 reasoning model, running on top of Ollama and OpenWebUI, all free and open-source software. You can install other models yourself through Ollama. As with any AI-generated output, verify it yourself before relying on it.
What DeepSeek's R1 model is
DeepSeek's R1 model is a specialised large language model designed for advanced chain-of-thought reasoning. It's built to handle complex questions and tasks more accurately by "thinking through" problems step by step, rather than producing a single-shot answer.
What ships on your server
Your server comes equipped with the 14B DeepSeek R1 reasoning model, offering a balance between performance and resource efficiency. It runs on top of Ollama and OpenWebUI. All three components, DeepSeek, Ollama and OpenWebUI, are free and open-source software: Worldstream provides the dedicated server the stack runs on, not the software itself. For more on each project, see their official sites:
Running a different model
The server ships with the 14B R1 build. To run something else, pull and run the model yourself with Ollama on your server. Check that the model fits the server's memory and GPU before you pull it.
Is it ready for production use?
As with any self-hosted LLM stack, you should verify and validate outputs before relying on them. Large language models can produce errors or hallucinations, so treat DeepSeek's output as a starting point rather than a final answer for anything that matters.