How We Review
Every score on this site comes from the same scorecard, applied the same way to every service. This page publishes it so that a number like “9.8” means something you can check. It was last applied on 27 September 2026.
Where the evidence comes from
We read what each service publishes: pricing and plan pages, feature pages, help-centre articles, terms of service, privacy notices and content policies. We record the date we read each one. We do not create accounts, pay for plans or enter credentials to gather evidence, so anything that appears only after signup — a price, a credit cost, a free-tier limit — is marked as not verifiable rather than guessed. Where a service’s own pages contradict each other, we report both statements and score the uncertainty.
We say “compared” and “reviewed”, not “tested”, because that is what this method is. We do not judge how well a character writes over a long session; that cannot be measured from public pages.
The eight criteria and their weights
Each criterion is scored from 0 to 10. The weighted score is the sum of each score times its weight, rounded to one decimal.
| Criterion | Weight | What it measures | A 9–10 looks like | A 7 or below looks like |
|---|---|---|---|---|
| Character builder depth | 20% | How many attributes you can set when you build a character from scratch — face, body, ethnicity, hair, outfit, personality, voice, relationship — and whether a free-text description is accepted on top of the presets. | Menus for face, body, ethnicity, hair and outfit plus personality, voice and relationship, and a free-text mode on top; male and trans options. | Presets you can adjust, or menus without personality and voice, or a builder that is announced but not live. |
| Image generation | 15% | Whether explicit images are allowed for adults, how the request is made (menu, prompt or in-chat), and which extra controls exist: editing, enhancing, group scenes, styles. | Explicit images on request in chat and in a generator, with editing, enhancing or multi-character scenes. | Images only on the top tier, or a single request path with no controls. |
| Video and voice | 15% | Published video length and modes, voice messages, live calls, and how each is billed. | Video of useful length, voice notes and live calls, each with a stated price. | Short clips only, or voice without calls, or media whose availability we could not confirm. |
| Memory and roleplay controls | 15% | What the service says about memory and whether you can adjust it, plus scenario, group-chat and roleplay tooling. | Memory you can size or extend, plus scenario and group tools. | Memory claimed with no controls, or explicitly limited. |
| Pricing transparency | 10% | Whether plan prices and per-action credit costs are readable before you sign up. | Plan prices and per-action credit costs readable before signup. | Plan prices hidden until after signup, and no per-action costs anywhere public. |
| Free access | 10% | What a new account can actually do before paying, stated in the service’s own words. | Meaningful free chat plus some credits to test images. | Paid per message, or a trial whose scope is not stated. |
| Privacy and deletion | 10% | Retention periods, model-training use of chats, and whether you can delete your account and history yourself. | Self-serve deletion, stated short retention, no training on private chats. | Deletion by request only, long retention, or claims that contradict the policy. |
| Content rules and safety | 5% | A public, specific prohibited-content policy, age checks, and a ban on real-person likeness. | A public, specific prohibited-content policy, age checks and a ban on real-person likeness. | Policy not publicly readable, or tags and rules that contradict each other. |
Why these weights
The first four criteria — builder, images, video and voice, memory — make up 65% because they decide what the experience is like. Pricing transparency, free access and privacy make up 30% because they decide whether you can trust the bill and the handling of your data. Content rules count 5%: every service we score meets the legal minimum, and the criterion separates clear, specific policies from vague ones.
Current scores
- LoveScape: 9.8 / 10 — full LoveScape review
- OurDreamAI: 9.6 / 10 — full OurDreamAI review
- CandyAI: 9.3 / 10 — full CandyAI review
- Joi AI: 8.7 / 10 — full Joi AI review
- GoLove: 8.4 / 10 — full GoLove review
What we could not verify
As of 2026-09: LoveScape and GoLove plan prices and credit costs (shown only after signup or behind an age and terms gate); CandyAI’s per-token costs and free-trial scope; GoLove’s real free message limit; OurDreamAI’s coin-pack prices; Joi AI’s Premium and Neuron-pack prices; and LoveScape’s full terms and privacy text. These gaps are reflected in the pricing-transparency and free-access scores rather than filled with estimates.
Updates and corrections
We re-read every service’s public pages monthly and change a score only when the evidence changes. If you spot an error, write to contact@ai-slut-generator.org with the page and the source; we correct confirmed mistakes and update the date on the page.