JevMade hello@JevMade.com
← Back to guides

JevMade field notes / API tutorial and benchmark

Use the Solar AI model with the Jev choice system

This guide explains how to use the Solar AI model as a stand-in for Jev. You will learn how to send it multiple-choice questions and get answers back in the exact same format Jev uses.

Original by Sung KimIntegrations

Listen to this guide

JevMade’s plain-English explanation

0:00 /

AI narration

Credits

“Wrap Solar Pro4 in a Jev-compatible decision API” by Sung Kim. Read the original source.

This expanded guide is an AI-narrated adaptation prepared by JevMade. It expands the source’s essential ideas, examples and caveats in JevMade’s own words and is not a word-for-word reading. The synthetic voice does not imitate the author or imply their endorsement.

Our summary

Jev is an AI tool that chooses from options rather than writing an open answer. This project is a translator that lets you use a different AI model, called Solar, in Jev's place. People use this to test Solar without changing the software they already built for Jev.

You send the translator a situation and a list of choices, along with a private access code for the Solar service. The software passes your question to Solar, asks it to pick a single letter, and then packages that choice to look exactly like a normal Jev response.

This setup is useful for developers comparing different AI tools. However, making Solar look like Jev does not mean it makes the same quality of decisions. The system relies on outside services to run, and the author's test scores are just a snapshot of how it performed once.

Key takeaways

  1. Send your questions to the special web address instead of a normal chat service.
  2. Use the exact same choice and scoring formats that you would normally use for Jev.
  3. Test the software yourself, because matching the format does not mean the answers will be perfect.

The test results are just snapshots reported by the author. A translated AI assistant is not the same as Jev and does not share its training.

GitHub README and reference implementation · Source reviewed

Read the original guide Opens the author’s site in a new tab.