Subscore

VarietyTesting methodology

Variety measures how much real choice the platform’s ready-made character library gives you.

A library can have hundreds of characters and still feel repetitive when most of them use the same styles, personalities, and scenarios. We look at the total amount of characters and how different those characters actually are.

6 evidence groups
11 scored tests

Variety is organized into 6 evidence groups. Some groups contain more than one scored test, which is why Variety has 11 scored tests in total.

How the Variety score is calculated

Every scored test is converted to a score from 0 to 10.

Not all evidence groups affect the Variety score equally. The weighted scores are combined to get the final Variety score. Variety makes up 34% of the Characters score.

30%Amount
15%Styles
20%Genders
10%Ethnicities
15%Personalities
10%Scenarios

Weighted evidence groups (combined 100%)Variety score34% of Characters score

View exact calculation

Each scored test gets a score from 0 to 10. Some scored tests count more than others. We apply the weights below and combine all the results to calculate the final Variety score.

Variety makes up 34% of the Characters score. Characters makes up 10% of the overall performance score.

Scored tests and weights

Scored test	How much it counts	See scoring
Female characters	13.33%	View
Male characters	5.19%	View
Anime female	13.33%	View
Anime male	5.19%	View
Styles	12.59%	View
Ethnicities	12.59%	View
Personalities	11.85%	View
Transgender	8.15%	View
Non-binary	2.96%	View
Other	2.96%	View
Scenarios	11.85%	View
Total	100%

How the score is calculated

Each test gets a score from 0–10
Test score × how much it counts
Example: 8 × 13.33% = 1.07 points
We do this for every test
We add all the points together
Final Variety score
Counts for 34% of Characters

Example calculation

We multiply each test score by how much it counts. We then add all the points together.

Scored test	Test score	How much it counts	Calculation	Points added
Female characters	8.00	13.33%	8.00 × 13.33%	1.07
Styles	6.00	12.59%	6.00 × 12.59%	0.76
Male characters	4.00	5.19%	4.00 × 5.19%	0.21
… other scored tests	—	—	Calculated same way	5.32
Final Variety score	—	100%	Add all points	7.35/10

Special cases

Not Applicable

If a scored test does not apply, we remove it and spread its weight equally across the remaining scored tests.

Unknown

If we cannot verify a result, the scored test receives a score of 0.

Manual adjustment

In rare cases, we may adjust a score when the calculated result is clearly unfair. We always record the reason.

Why Variety matters

Having a big character library matters because people eventually run out of interesting characters to talk to.

This is especially important when you pay for an annual plan. An app might look exciting during the first week, but it can become boring quickly when the same few personalities, looks, and roleplay ideas keep appearing.

Variety is not only about the total number. A platform with 200 nearly identical characters may offer less real choice than a platform with 80 characters spread across different styles, genders, personalities, and scenarios.

How we test Variety

We use a paid account and open the full ready-made character library available during testing.

We record the test date, the account settings used, and whether adult content or other library filters were enabled.

When the platform shows a reliable total, we record it. When no total is shown, we count the visible listings where possible. If the library cannot be counted reliably because of endless loading, hidden pages, or changing results, we report the limitation.

We only use categories and labels shown by the platform. We never guess a character’s gender or ethnicity from an image.

Duplicate characters are still included in the library count. We test duplicates separately under Character Quality, so the same problem is not judged twice.

Limitations and edge cases

Character libraries change frequently. New characters may be added, old profiles may be removed, and categories may be reorganized after we finish testing.

Some libraries use endless scrolling or personalized recommendations, which can make exact counts difficult. We report Unknown or explain the limitation when a reliable count is not possible.

Access may also depend on your subscription, region, or adult-content settings. Our results reflect the paid test account and settings used on the recorded test date.

We only use labels provided by the platform for gender and ethnicity. We never guess these details from a character’s appearance.

Evidence groups

Variety has 6 evidence groups made up of 11 scored tests.

Amount measures how many ready-made characters are available across the main character types.

We score Female Characters, Male Characters, Anime Female Characters, and Anime Male Characters separately because the size of each group can be very different.

A small library can become repetitive quickly. Looking at each main character type separately also prevents one very large group from hiding the fact that other users have almost no choice.

Scored test	How much it counts	View
Female characters	10.80%	View
Male characters	4.20%	View
Anime female characters	10.80%	View
Anime male characters	4.20%	View

These percentages add up to the 30% Amount weight above.

Female characters

10.80%of Amount (30% total)

Counts the ready-made female characters available in the library.

How we test

We count every character clearly labelled as female by the platform. We never guess a character’s gender from their appearance.

What counts

Female characters shown in the ready-made library
Female characters available through the paid test account

What does not count

Gender guessed from an image
Private characters
Characters created by users
Duplicate listings that point to the same profile

Result shown

82 female characters

Scoring

More characters means a higher score. We use the ranges below.

Characters	Score
10 or fewer	2/10
11–30	4/10
31–80	6/10
81–120	8/10
121+	10/10

A result of 82 female characters scores 8/10.

Male characters

4.20%of Amount (30% total)

Counts the ready-made male characters available in the library.

How we test

We count every character clearly labelled as male by the platform. We never guess a character’s gender from their appearance.

What counts

Male characters shown in the ready-made library
Male characters available through the paid test account

What does not count

Gender guessed from an image
Private characters
Characters created by users
Unlabelled characters

Result shown

21 male characters

Scoring

More characters means a higher score. We use the ranges below.

Characters	Score
5 or fewer	2/10
6–15	4/10
16–30	6/10
31–50	8/10
51+	10/10

A result of 21 male characters scores 6/10.

Anime female characters

10.80%of Amount (30% total)

Counts the ready-made anime-style female characters available in the library.

How we test

We count female characters that clearly use anime-style art, using platform labels and style categories where available.

What counts

Anime-style female characters
Characters listed under anime or 2D art categories when gender is also labelled

What does not count

Realistic characters with only minor cartoon styling
Promotional images that are not actual character profiles
Gender guessed from an image

Result shown

28 anime female characters

Scoring

More characters means a higher score. We use the ranges below.

Characters	Score
10 or fewer	2/10
11–30	4/10
31–80	6/10
81–120	8/10
121+	10/10

A result of 28 anime female characters scores 4/10.

Anime male characters

4.20%of Amount (30% total)

Counts the ready-made anime-style male characters available in the library.

How we test

We count male characters that clearly use anime-style art, using the same style rules as anime female characters.

What counts

Anime-style male characters
Characters listed under anime or 2D art categories when gender is also labelled

What does not count

Realistic male characters
Style inferred without a clear platform label
Gender guessed from an image

Result shown

6 anime male characters

Scoring

More characters means a higher score. We use the ranges below.

Characters	Score
5 or fewer	2/10
6–15	4/10
16–30	6/10
31–50	8/10
51+	10/10

A result of 6 anime male characters scores 4/10.

Evidence group 2 of 6

Styles

15%of Styles (15% total)

How many clearly different visual styles are available.

A library can be large but still feel limited when every character has the same realistic look. Different styles give users more ways to explore the app.

How we test

We review the library and count clearly different styles such as realistic, anime, 2D cartoon, 3D render, and fantasy.

What counts

Realistic
Anime
2D or cartoon
3D render
Fantasy
Other clearly different styles

What does not count

Minor lighting changes
Image-quality settings
Filters that do not change the actual art style
Two labels that produce almost identical results

Result shown

4 styles

Realistic, Anime, 2D Cartoon and Fantasy

Scoring

More styles means a higher score. We use the ranges below.

Styles	Score
1 or fewer	2/10
2	4/10
3	6/10
4	8/10
5+	10/10

A result of 4 styles scores 8/10.

Evidence group 3 of 6

Transgender characters

11.58%of Genders (20% total)

The number of ready-made transgender characters in the library.

Transgender representation matters for users who specifically want that type of companion. It also signals whether the platform treats gender diversity as a real browsing option.

How we test

We count characters explicitly labelled as transgender by the platform. We never infer this label from appearance.

What counts

Transgender characters with explicit platform labels

What does not count

Gender inferred from an image
Characters only described indirectly in profile text
Duplicate listings

Result shown

7 transgender characters

Scoring

More characters means a higher score. We use the ranges below.

Characters	Score
0 or fewer	0/10
1–3	4/10
4–10	6/10
11–25	8/10
26+	10/10

A result of 7 transgender characters scores 6/10.

Non-binary characters

4.21%of Genders (20% total)

The number of ready-made non-binary characters in the library.

Non-binary characters give users another meaningful choice beyond binary gender categories. Very small counts can still be useful when the platform clearly supports the label.

How we test

We count characters explicitly labelled as non-binary by the platform.

What counts

Non-binary characters with explicit platform labels

What does not count

Gender inferred from an image
Unlabelled characters
Duplicate listings

Result shown

3 non-binary characters

Scoring

More characters means a higher score. We use the ranges below.

Characters	Score
0 or fewer	0/10
1–2	4/10
3–5	6/10
6–15	8/10
16+	10/10

A result of 3 non-binary characters scores 6/10.

Other gender groups

4.21%of Genders (20% total)

The number of ready-made characters in other explicitly labelled gender groups not covered elsewhere.

Some platforms use additional gender labels beyond female, male, transgender, and non-binary. This test captures those options when they are clearly offered.

How we test

We count characters in any additional gender groups explicitly named by the platform.

What counts

Other explicitly labelled gender groups shown by the platform

What does not count

Groups already counted under female, male, transgender, or non-binary
Gender inferred from appearance
Vague or duplicate labels

Result shown

2 characters in other gender groups

Scoring

More characters means a higher score. We use the ranges below.

Characters	Score
0 or fewer	0/10
1–2	4/10
3–5	6/10
6–15	8/10
16+	10/10

A result of 2 characters in other gender groups scores 4/10.

Evidence group 4 of 6

Ethnicities

10%of Ethnicities (10% total)

The number of ethnicity categories explicitly shown by the platform.

A wider selection makes it easier for users to find characters who match their preferences or the type of relationship they want to explore.

How we test

We count the ethnicity categories shown in the library, filters, or character labels.

What counts

Explicit platform categories
Clearly named ethnicity filters
Categories that lead to different groups of characters

What does not count

Ethnicity guessed from appearance
Country labels unless the platform clearly uses them as ethnicity categories
Two labels that clearly mean the same thing

Result shown

12 ethnicity categories

Scoring

More categories means a higher score. We use the ranges below.

Categories	Score
3 or fewer	1/10
4–5	2/10
6–7	3/10
8–10	5/10
11–15	7/10
16+	10/10

A result of 12 ethnicity categories scores 7/10.

Evidence group 5 of 6

Personalities

15%of Personalities (15% total)

How many distinct personality types users can browse or filter by.

Different pictures do not mean much when every character talks and behaves the same way. Personality variety helps the library feel less repetitive.

How we test

We count personality categories and filters shown in the character library.

What counts

Clearly different personality categories
Personality filters that lead to meaningfully different characters
Platform-labelled personality types

What does not count

Near-identical labels such as “shy” and “slightly shy” when they clearly mean the same thing
Personality details hidden inside individual profile descriptions
User-created personality prompts

Result shown

18 personality types

Scoring

More personality types means a higher score. We use the ranges below.

Types	Score
2 or fewer	2/10
3–5	4/10
6–10	6/10
11–20	8/10
21+	10/10

A result of 18 personality types scores 8/10.

Evidence group 6 of 6

Scenarios

10%of Scenarios (10% total)

Checks if there are different roleplay scenarios or chat modes besides the regular chat.

Scenarios give users different reasons to start a conversation. Without them, a large library can still feel like the same basic chat repeated with different profile pictures.

How we test

We counted all chat modes and roleplay scenarios users can select. This includes both the character library and the custom AI character section.

What counts

Girlfriend or boyfriend
Friend or spouse
Stepsibling or roommate
Fantasy roles
Workplace roles
Dominant or submissive scenarios
Other meaningfully different relationship or story types

What does not count

Small wording changes to the same scenario
Individual opening messages
Scenarios only available through user-created prompts
Tags that do not change the relationship or story

Result shown

14 scenarios

Scoring

More scenarios means a higher score. We use the ranges below.

Scenarios	Score
2 or fewer	2/10
3–5	4/10
6–10	6/10
11–20	8/10
21+	10/10

A result of 14 scenarios scores 8/10.

Back to Characters