Chat (IDE)

Inline completion

Ghost-text completions that suggest the next line as you type — powered by the same bring-your-own-key providers as chat, and included on every plan.

How it works

As you type in the editor, Tempr shows a faded ("ghost text") suggestion for what comes next, right at your caret. Take it, keep typing to ignore it, or just keep going — it never interrupts you with a panel or a dialog.

  • Same keys, same catalog — completions run on your own provider key, on any model your plan entitles you to.
  • Inline-only — completion is a separate surface from the chat panel; using one doesn't open the other.
  • Every plan — Standard, Pro, and Teams all include full inline completion.

A model per scenario

Not every keystroke deserves the same model. Tempr lets you tune the completion model per scenario, so you can pair a fast, cheap model for routine next-line suggestions with something heavier for chat and agent work — all on the same key.

ScenarioSuggestion
Everyday next-line completionA fast, inexpensive model — latency matters more than depth here.
Chat & agent turnsYour strongest model — reasoning quality pays off.
Boilerplate-heavy filesWhatever you use for chat; repetitive code is where suggestions shine.

Set your completion model in Tempr's settings, alongside your chat default.

Tips

  • Completions ground on what's visible in the editor — an open file with a few examples of the pattern you want produces dramatically better suggestions than a bare file.
  • Because suggestions run on your own key, completion volume shows up in your provider usage — pairing a cheap completion model with a premium chat model keeps costs sane.
  • If suggestions feel slow, switch the completion scenario to a faster model before blaming your connection — it's usually the model, not the pipe.
Note

Inline completion is available in both Visual Studio (2022 and 2026) and VS Code. Looking for whole-task generation instead of next-line suggestions? That's agent mode.