I was digging through Continue's settings after the latest update and hit the "Autocomplete Provider" dropdown. The options are "OpenAI" and "Ollama," but the description is pretty sparse. I get that it's for code completion, but I'm trying to understand the *scope*.
Does this control *all* inline suggestions? For example:
- When I'm writing a new Terraform `resource` block, does this setting dictate which model finishes it?
- What about when I'm halfway through a `local.module` reference—does it try to suggest the correct outputs?
- If I pick Ollama, does it only use my local model for these quick tab-completions, while the main chat could still be using GPT-4?
I'm mostly an Ollama user for local work, but I'm wary of it slowing down my editor if it's firing on every keystroke. Conversely, if I set it to OpenAI, I'm worried about cost or API latency during regular typing.
Has anyone mapped this out? A concrete example of the difference in behavior would be super helpful. My main concern is keeping my Terraform module development efficient without spinning up a costly or slow completion process.