LMpad
6 min read · Site tutorial

Variables, models, streaming, and saved outputs

How to get reliable results from the run panel — free tier, your own key, and saving outputs.

  1. 1

    Open the run panel

    From a note page click Run prompt, or from My Pad open Run from the card menu. The run view opens inline on your pad so the board stays one click away.

    Step 1: Open the run panel
  2. 2

    Fill {{variables}}

    Templates with placeholders like {{topic}} or {{audience}} show input fields. The run button stays disabled until required variables are filled — this prevents half-baked prompts.

    Step 2: Fill {{variables}}
  3. 3

    Pick free or paid tier

    Toggle between free models (Llama, Qwen, Gemma, etc.) and paid models (GPT, Claude, Gemini). Paid models require your OpenRouter key in Settings.

    Open Settings
    Step 3: Pick free or paid tier
  4. 4

    Stream and review

    Output streams live in rendered Markdown or raw text. Stop mid-stream if the answer drifts. Run again after tweaking variables or the prompt itself.

    Step 4: Stream and review
  5. 5

    Save outputs to the note

    On pad notes, save outputs under the note so you build a history of runs — useful for comparing models or iterating on drafts.

    Step 5: Save outputs to the note

Tips

  • Run the same prompt on two models using the model comparison template in Guides.
  • Shorten prompts that ramble — move long context into variables instead of the fixed body.
  • If a run fails, check Settings for a valid key or switch to a free model.

Continue learning