How we review AI girlfriend apps
Every app we cover is scored out of 10 using the same five-part rubric. We pay for the app ourselves, use it like a real user for days, and only then assign a score. Here's exactly what goes into it.
1. Conversation & memory
30%Does it stay in character? Remember earlier messages? Hold a long conversation without looping or breaking? We run multi-day chats, not five-minute demos.
2. Images & media
25%Image quality, how consistent your companion's face stays across generations, and how generous the app is with requests on each tier.
3. Voice & extras
15%Voice message and call quality, plus any standout features like custom personalities, scenarios, or phone calls.
4. Value for money
15%What you actually get free vs paid, where the real paywall sits, and whether the price matches the experience. We flag pricing traps.
5. Privacy & safety
15%How the app handles your data, what its policy says, billing discretion, and content guardrails. We read the privacy policy so you don't have to.
How we score and re-test
Scores are weighted by the percentages above and rounded to one decimal. Apps change constantly, so we re-test the ones we cover and update the “last updated” date on each review when we do. A higher score means a better all-round experience for most people, but our category rankings exist precisely because the best app for you depends on what you want.
Independence
We earn affiliate commissions, but testing and scoring happen before any commercial consideration. We never let a partnership change a score, and we publish apps' weaknesses plainly. Read our editorial policy and affiliate disclosure for the full picture.
Use of AI in our content
We may use AI tools to help draft and structure articles, but every page is based on real hands-on testing and is reviewed and edited by a named human editor before it's published. We don't publish untested, machine-generated reviews.