Self-hosted via Ollama or using the custodial version of deepseek?
Self hosted. Been tinkering with LM Studio for hosting and managing models locally.