Best AI chatbot for roleplay, ranked by what users actually think
One score per service, refreshed twice a day from a rolling 30-day window of Reddit posts and app-store reviews. Every number traces back to real posts — no vendor benchmarks, no ad money in the ranking.
RedditStoresMemoryBarometerSources and the barometer run 0–100; memory is a percentage. Each source is scored from labeled items; the smaller figure under a score is how many items it rests on. Rank is by barometer and stays with a service when you re-sort.
Gauge:under 4040–55over 55
Filters by price, content policy and platform arrive with the per-service attributes; those are not in the data yet.
6measured·0listed without a score
Below this line our evidence runs out
Every service we track cleared the evidence bar this window. When one does not, it stays on this page — without a number.
What this window rests on
- Services measured
- 6
- Sources counted
- 11
- Posts and reviews labeled
- 6,004
weights: reddit .35 / stores .25 (renormalized) · how the score works →