Skip to content

Choice of LLMs

One of the first questions operators ask is: “How much is this going to cost me?”

The answer: it’s entirely up to you.

ADIF-MCP doesn’t ship with, or require, a specific Large Language Model (LLM). Instead, it provides a standard interface so you (or developers) can connect whatever model fits your needs, budget, and shack setup:

  • Free / Local Models
  • Run open-source LLMs (like LLaMA-3, Mistral, or Mixtral) on your own PC, NUC, or server.
  • Tools like Ollama make this easy to run offline, with zero cloud cost.

  • Great for experimenters, DIYers, or anyone who prefers not to rely on the internet.

  • Commercial Cloud Models

  • Use providers like OpenAI (ChatGPT), Anthropic (Claude), or Google (Gemini).
  • These are “plug and play” with powerful reasoning and large context windows.

  • Costs vary, but many have free tiers and low-cost options for light use.

  • Hybrid Setups

  • Some operators may use a small free local model for day-to-day queries, and keep a stronger paid model in reserve for heavy lifting (e.g., log analysis or award prep).

In short

  • Operators can choose free local if cost is the priority.
  • Developers can choose best-in-class paid models if capability is the priority.
  • Everyone gets the same MCP backbone — the choice of LLM is up to you.

Just like antennas or radios, there’s no one “right” answer: your shack, your choice.