All bots

How we test

Trending tells you what people are sharing on X. Our score tells you whether the bot did the job. We never guess a score.

We use 2 to 3 prompts for each thing the bot says it does, plus a safety check.

Our score

The score is a whole number out of 100. It shows how well the bot did the job it was made for in our test. We don't sort scores into grades or labels, so compare the numbers directly. A higher score means the bot did more of its job well.

If safety is under 5 out of 10, the score is at most 59. Bots made by the xAI team are held and not scored.

A published score is final. Until then the bot shows a grey Still testing badge and no number. That includes rankings, categories, Trending, the bot page, share cards, and search. Sub-scores stay off the page until the score is final. Rankings list final scores only, highest to lowest.

Best for

Each tested bot also gets a Best for line.

Peace of mind

Every tested bot gets a plain note on setup, accounts, what it reads, whether it sends data out, whether it acts without asking, plan usage, how to stop or undo it, and how it handles mistakes. That note is not a score.

How it could score higher

A bot with a final score can include one to three specific fixes from the test, with an estimate of the score it would reach if those were fixed. The estimate is written by the tester. The page does not invent one, and it stays hidden while the bot is still testing.

Resubmits

Makers can send an update from the submit page. We rerun the full test. Resubmits are limited to once per bot every 30 days. The old score and the new score stay in the score history. Each rescore gets fresh tips. Paying never buys a better score.

Our rules