Quickstart: ask your first question

You can try Keplar without signing up. An account adds saved chats, memory and the higher plans, but the first question needs nothing.

Steps

1. Open the app. The composer is the box at the bottom of the screen.
2. Type a question that has more than one reasonable answer, for example "Should a two-person startup rewrite a working Node.js backend in Rust before it has paying customers?" Questions with a trade-off show the product best.
3. Press Send. Keplar first shows which kinds of capability the question needs, then which models it chose, then the answer as it is written.
4. Read the short answer at the top. Then open the sections underneath: Consensus, Models consulted, Sources, Disagreements, Verification and Reasoning summary.
5. Ask a follow-up in the same chat, such as "What would change your mind?" The recent messages of the chat travel with your next question, so follow-ups work without repeating yourself.
6. Open the options next to the composer and try Most thorough on the same question to see a wider panel.

What to expect on the Free plan

Free runs on free open models. It is real multi-model Keplar, but those models can be slower when busy, and their hosts may log prompts, so do not paste anything sensitive. Free allows up to 50 questions a day under fair use. If a model is slow, Keplar may start a stand-in from another family and will say so under the answer.

Things that are not a problem

• A one-model answer. Short, well-defined questions go to a single fast model. That is by design; see Which questions use more models.
• "2 of 3 models answered." A model that times out or declines is not counted as disagreement. The answer says who was missing and offers a retry.
• No disagreement section. It appears only when the models took more than one position. Its absence means the models that responded were aligned, not that the answer is correct.

Next

Learn the layout in Read a Keplar answer, then improve your questions with Write better questions.

Related

• Read a Keplar answer: A tour of the six sections under every multi-model answer, what each one means, and what it does not mean.
• Write better questions: Eight habits that get more from a multi-model answer: state the decision, add constraints, ask for the counterargument, separate facts from judgment, and more, with prompts you can copy.
• Thoroughness modes: Automatic, Fastest, Balanced and Most thorough: what each changes about panel size and the judge, when to use each, and how they interact with your plan.