Research feed

Litmiq Labs

On this page

The short answer

  • Labs is a feed of short, dated, reproducible prompt-eval notes.
  • The first post is the method. There are no findings on this page yet.
  • We will not invent a score, a ranking or a “winner” to fill the feed.

Feed

Findings

No findings published yet. The next Labs note will appear here after a dated, reproducible run is completed.

If you are looking for already published research, use the reports hub and the reliability benchmark.

What we publish

A Labs note will be a real run or a method change. It will be dated. It will name the suite, the models, the judge and the parameters, or it will say what is still unknown. It will not invent testimonials, user counts or unofficial rankings.

Model providers who want a launch-day run should read the provider offer. The public table lives on the leaderboard, which is also pending its first run.

FAQ

Where are the findings?

There are none yet. The first item in this feed is the method. A findings note will be dated and will link to a real run. We will not invent a result to fill this page.

How is this different from the reports hub?

Reports already hosts original research such as the reliability benchmark and the laws and pricing studies. Labs is the feed for short, dated eval runs and method notes.

Can I reproduce a future run?

Yes, once a run is published. Each note will link to the suite and the parameters. The method page is the standing rulebook.

Read the method

The rules are public before any number is.

  • Free, no signup
  • Runs in your browser
  • Your OpenAI key, sent only to OpenAI