RealTimeX Local

RealTimeX Local is the managed built-in local LLM provider in RealTimeX.

Use it when you want RealTimeX to run GGUF chat models on your own machine without connecting to a separate Ollama, LM Studio, or other external server.

Where it appears

• provider selection: Settings > AI Providers > LLM
• runtime and built-in model management: Settings > AI Providers > RealTimeX Local
• broader cross-provider inventory: Settings > AI Providers > Local Model Management

What this provider controls

Choosing RealTimeX Local in the LLM selector tells RealTimeX to use the managed built-in local runtime as the system LLM provider.

That selection depends on a second setup layer:

• the runtime must be installed
• a hardware backend must be available
• at least one local model must be downloaded
• one model must be selected as the preferred built-in model

Current setup flow

1. Open Settings > AI Providers > LLM.
2. Choose RealTimeX Local.
3. Open Settings > AI Providers > RealTimeX Local.
4. Download the runtime if it is missing.
5. Keep Autodetect for the backend unless you have a reason to pin Metal, CUDA, Vulkan, or CPU only.
6. Download a recommended GGUF model, search Hugging Face, or enter a repository or file path manually.
7. Set the model as the preferred built-in model and load it.
8. Return to chat and test a normal conversation.

What the current runtime page manages

The current RealTimeX Local page is more than a simple model dropdown.

It can manage:

• runtime installation state
• backend selection
• context size
• model download and load state
• warmup readiness
• unload, refresh, delete, and repair actions

If you are looking for those controls, use the dedicated page instead of expecting everything inside the LLM selector itself.

When to use this provider

• Use RealTimeX Local when you want the simplest built-in local inference flow.
• Use it when you do not want to run a separate local model server.
• Use it when on-device privacy or offline-capable chat matters more than cloud-only convenience.

If you want to connect RealTimeX to an external local provider instead, use pages such as Ollama, LM Studio, or llama.cpp.

Related guides

• RealTimeX Local
• Local Models
• Large Language Models