How we test
Each platform gets scored across five areas on a ten-point scale. The star rating is just the average, rounded to the nearest half star. Same categories. Same weights. Every single app.
The categories
| Chat quality | Whether it holds a conversation and stays in character, and whether it remembers you between sessions. |
| Images & video | Quality, speed and whether generated pictures look like the character; video where offered. |
| Voice | Voice notes, calls and read-aloud: quality, latency and consistency. |
| Customisation | How much of the personality, appearance and backstory you control and how well it sticks. |
| Value | What the paid tier includes against its price, and how limited the free tier is. |
The protocol
The reviewer buys a paid account on each app. Runs the same three character builds. Follows a scripted week: small talk, a set roleplay scene, a memory check at the 48-hour mark, image requests from identical prompts, plus a voice call if the platform supports it. Ratings made before that testing draws on documented features, current pricing, and published reviews. They get revised once the protocol runs.
What we exclude
Any platform that allows content with minors or non-consensual images of real people. No platform buys its rating, its placement, or its mention.