search

Subscore

QualityTesting methodology

Quality measures how original, complete, and visually polished the ready-made character library feels.

A platform can have hundreds of characters and still feel low quality. Some apps fill their library with copied profiles, weak descriptions, and broken images just to make the total number look more impressive.

We review one fixed sample of characters and use the same sample for every Quality test.

  • 4 evidence groups
  • 4 scored tests

Quality is organized into 4 evidence groups. Each group contains one scored test: Duplicates, Originality, Profile Quality, and Visual Quality. All four tests count equally toward the final Quality score.

How the Quality score is calculated

Every scored test gets a score from 0 to 10.

All four tests count equally. We combine their scores to calculate the final Quality score. Quality makes up of the Characters score.

  • 25%Duplicates
  • 25%Originality
  • 25%Profile Quality
  • 25%Visual Quality
Weighted evidence groups (combined 100%)Quality score of Characters score
View exact calculation

Each test gets a score from 0 to 10. We multiply every score by 25% and add the points together.

Quality makes up of the Characters score. Characters makes up 10% of the overall performance score.

Scored tests and weights

Scored testHow much it countsSee scoring
Duplicates25.00%View
Originality25.00%View
Profile Quality25.00%View
Visual Quality25.00%View
Total100%

How the score is calculated

  1. Each test gets a score from 0–10

  2. Test score × how much it counts

    Example: 9.40 × 25.00% = 2.35 points

  3. We do this for every test

  4. We add all the points together

  5. Final Quality score

    Counts for 33% of Characters

Example calculation

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

Scored testTest scoreHow much it countsCalculationPoints added
Duplicates9.4025.00%9.40 × 25.00%2.35
Originality8.2025.00%8.20 × 25.00%2.05
Profile Quality8.4025.00%8.40 × 25.00%2.10
Visual Quality9.0025.00%9.00 × 25.00%2.25
Final Quality score—100%Add all points8.75/10

Special cases

Not Applicable

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

Unknown

If we cannot verify a result, the 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 Quality matters

A large character library does not automatically mean a good character library.

Some platforms add new characters as quickly as possible to make their library look bigger. You often end up with copied personalities, nearly identical images, empty profiles, and the same basic scenario repeated with a different name.

That can make the app feel boring even when it claims to offer hundreds of characters.

Strong character quality means the profiles feel different, give you enough information before starting a chat, and use images that look clean and well made.

That is why we check the library itself instead of only counting how many characters are available.

How we test Quality

We use a paid account and choose one fixed sample from the ready-made character library.

We use the same characters for all four Quality tests. This prevents us from using the strongest profiles for one test and weaker profiles for another.

First, we check the sample for duplicate or near-duplicate profiles.

We then check whether the characters have different appearances, personalities, and scenarios.

Next, we review how complete each profile is.

Finally, we review the main profile image for clarity, visual problems, and overall presentation.

Duplicates still count toward the library totals under Variety. Quality is where we judge whether those listings are actually different and well made.

Limitations and edge cases

Quality is based on a fixed sample, not every character in the library. The sample helps us spot clear patterns, but it cannot guarantee that every profile has the same level of quality.

Character libraries also change regularly. New profiles may be added and weak profiles may be improved or removed after testing.

Originality and visual quality can involve some judgement. We reduce this by using the same checklist for every platform.

Public community characters may have very different quality from characters made by the platform. We record which type of library was included in the sample.

Evidence groups

Quality has 4 evidence groups made up of 4 scored tests.

25%

Duplicates

1 scored test

Duplicates measures how many characters in the sample appear to be copies or near-copies of another profile.

A large number of duplicates can make a big library feel much smaller than it looks.

Duplicates

The percentage of reviewed profiles that repeat another character.

How we test

We review every character in the fixed sample and compare it with the other profiles. We count a profile as a duplicate when most of these parts are nearly the same: main image, name, description, personality, and scenario.

What counts
  • The same profile image with only a small change
  • Nearly identical names and descriptions
  • The same personality and scenario reused across profiles
  • Profiles that clearly look like copied versions of one another
What does not count
  • Characters that share one trait but are otherwise different
  • Characters from the same theme with different personalities
  • Small similarities that are common across the whole app
  • Profiles that use the same visual style but represent different characters
Result shown

3 duplicates in a sample of 50 profiles

Duplicate rate: 6%

Scoring

Fewer duplicates means a higher score.

Duplicate rateScore
0%10/10
20%8/10
40%6/10
60%4/10
80%2/10
100%0/10

A duplicate rate of 6% scores 9.4/10.

The exact score moves with the duplicate rate.

Evidence group 2 of 4

25%

Originality

1 scored test

Originality measures how different the characters feel from one another.

Changing the name or hair color is not enough when the personality and scenario are still almost identical.

Originality

The percentage of reviewed characters that feel original.

How we test

We review the same fixed sample. Each character receives one point for a distinct appearance, a distinct personality, and a distinct scenario. A character passes when it gets at least 2 out of 3 points.

What counts
  • A look that is clearly different from the other profiles
  • A personality that changes how the character behaves
  • A scenario that gives the conversation a different starting point
What does not count
  • A name change on an otherwise copied profile
  • A small color change to the same image
  • Rewording the same personality description
  • Reusing the same relationship or roleplay idea with minor changes
Result shown

41 of 50 characters passed

Originality rate: 82%

Scoring

A higher originality rate means a higher score.

Characters that passScore
20%2/10
40%4/10
60%6/10
80%8/10
100%10/10

A result of 82% scores 8.2/10.

The exact score matches the percentage.

Evidence group 3 of 4

25%

Profile Quality

1 scored test

Profile Quality measures whether character profiles give you enough useful information before you start chatting.

A good profile should explain who the character is and what kind of conversation or relationship to expect.

Profile Quality

How complete and useful the character profiles are.

How we test

We review the same fixed sample. Each profile receives one point for including a name, a clear description, personality information, relationship or scenario information, and example dialogue or an opening message. Each profile can receive up to five points.

What counts
  • Information clearly shown on the profile
  • A description that explains who the character is
  • Personality details that help set expectations
  • A clear relationship or scenario
  • An opening message or example of how the character talks
What does not count
  • Information that only appears after starting the chat
  • Empty or placeholder text
  • Repeated text copied across several profiles
  • Vague descriptions that do not explain the character
Result shown

210 of 250 profile checks passed

Profile Quality: 84%

Scoring

More completed profile checks means a higher score.

Checks passedScore
20%2/10
40%4/10
60%6/10
80%8/10
100%10/10

A result of 84% scores 8.4/10.

The exact score matches the percentage.

Evidence group 4 of 4

25%

Visual Quality

1 scored test

Visual Quality measures how clear and well made the main character images look.

A profile image is often the first thing you see. Broken faces, damaged bodies, or blurry images can make the whole library feel rushed.

Visual Quality

The quality and consistency of the main profile images.

How we test

We review the main image of every character in the same fixed sample. Each image receives one point for a clear face, a clear body, no major anatomy errors, no obvious image damage, and good overall presentation. Each image can receive up to five points.

What counts
  • The face is clear and easy to see
  • The body looks complete and believable
  • Hands, limbs, and clothing are not badly broken
  • The image is not blurry, corrupted, or heavily damaged
  • The image looks suitable for a finished character profile
What does not count
  • Small style choices that are clearly intentional
  • Minor background problems that do not affect the character
  • Personal preference for realistic or anime art
  • Image quality inside the chat or image generator
Result shown

225 of 250 image checks passed

Visual Quality: 90%

Scoring

More visual checks passed means a higher score.

Checks passedScore
20%2/10
40%4/10
60%6/10
80%8/10
100%10/10

A result of 90% scores 9.0/10.

The exact score matches the percentage.

Back to Characters