search

Subscore

MediaTesting methodology

Media measures what you and the AI character can send inside the chat.

A normal text chat can start to feel basic after a while. Images, voice messages, videos, GIFs, and reactions make the conversation feel more personal and much more immersive.

We do not only check whether an app advertises these features. We try each one three times to see whether it actually works.

  • 7 evidence groups
  • 7 scored tests

Media is organized into 7 evidence groups. Each group contains one scored test: Images Sent, Images Received, Voice Sent, Voice Received, Chat Video, GIFs, and Reactions.

How the Media score is calculated

Every test gets a score from 0 to 10.

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

The live methodology gives more weight to media you receive from the character than media you send yourself. Media makes up of the Chat Features score.

  • 7%Images Sent
  • 21%Images Received
  • 7%Voice Sent
  • 23%Voice Received
  • 20%Chat Video
  • 2%GIFs
  • 20%Reactions
Weighted evidence groups (combined 100%)Media score of Chat Features score
View exact calculation

Each test gets one of these scores: Yes — 10/10, Limited — 5/10, No — 0/10, or Unknown — 0/10. We multiply every score by how much it counts and add the points together.

Media makes up of the Chat Features score. Chat Features makes up 10% of the overall performance score.

Scored tests and weights

Scored testHow much it countsSee scoring
Images Sent7.00%View
Images Received21.00%View
Voice Sent7.00%View
Voice Received23.00%View
Chat Video20.00%View
GIFs2.00%View
Reactions20.00%View
Total100%

How the score is calculated

  1. Each test gets a score from 0–10

  2. Test score × how much it counts

    Example: 10 × 7.00% = 0.70 points

  3. We do this for every test

  4. We add all the points together

  5. Final Media score

    Counts for 30% of Chat Features

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
Images Sent10.007.00%10.00 × 7.00%0.70
Images Received5.0021.00%5.00 × 21.00%1.05
Voice Sent10.007.00%10.00 × 7.00%0.70
Voice Received10.0023.00%10.00 × 23.00%2.30
Chat Video0.0020.00%0.00 × 20.00%0.00
GIFs5.002.00%5.00 × 2.00%0.10
Reactions10.0020.00%10.00 × 20.00%2.00
Final Media score—100%Add all points6.85/10

Special cases

Not Applicable

If the app clearly does not support the underlying feature, the related test may be removed and its weight is spread across the remaining tests.

Unknown

If we cannot confirm whether a feature works, it 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 Media matters

Media makes an AI girlfriend chat feel like more than a normal chatbot.

Receiving a photo can make the conversation feel more visual. A voice message lets you hear the character’s personality instead of only reading it. Video can make the whole experience feel much more immersive.

The difference between apps can also be huge.

Some AI girlfriend apps let you send photos, receive voice replies, request videos, and react to messages without leaving the conversation. Others only offer basic text chat, even though their homepage makes the app look much more advanced.

These features also do not always work reliably. An app may claim that characters can send images, but the request fails twice out of three attempts. Voice replies may only work with certain characters, or video may be locked behind an extra payment.

That is why we try every feature instead of only checking whether there is a button for it.

How we test Media

We use a paid account and test every available media feature inside the chat.

We try each feature three times in separate conversations.

For media sent by the user, we upload or record three different files.

For media received from the character, we make the same request in three separate chats and record whether the character sends the requested result.

We record whether all three attempts worked, whether only some attempts worked, any important restrictions, whether credits were used, and whether the media appeared properly inside the conversation.

The current guided test includes Images Sent, Images Received, Voice Sent, Voice Received, Chat Video, GIFs, and Reactions.

Limitations and edge cases

Media features can be different depending on the character, device, or subscription.

A voice feature may work on the mobile app but not on desktop. Some characters may support images while others do not.

A feature may also technically work but still have important restrictions. For example, users may only be able to send certain file types, or the character may only send pre-made images instead of generating something based on the conversation.

We record these restrictions and use Limited when the feature exists but does not work properly in every test.

This score checks whether media can be exchanged inside the chat. The actual quality of generated images and videos is tested separately under Images and Video.

Media costs are also tested separately under Pricing.

Evidence groups

Media has 7 evidence groups made up of 7 scored tests.

7%

Images Sent

1 scored test

Images Sent measures whether you can send your own images to the AI character.

This can make the conversation more personal because you can show the character what you are talking about instead of trying to explain everything with text.

Images Sent

Whether you can send your own images to the AI character.

How we test

We send three different image files in three separate chats. We check whether every image uploads, appears in the conversation, and can be understood by the character.

What counts
  • Uploading an image from your device
  • Sending the image inside a normal chat
  • The image appearing properly in the conversation
  • The character responding to the image
What does not count
  • Uploading a profile picture
  • Adding an image during character creation
  • Sharing an outside link instead of the actual file
  • An upload button that repeatedly fails
  • Images that appear in chat but cannot be seen by the AI
Result shown

Yes — all 3 images were sent successfully

Scoring

We use the result categories below.

ResultScore
Yes — all three attempts worked10/10
Limited — only some attempts worked or restrictions were important5/10
No — none of the attempts worked0/10
Unknown0/10

In this example, Images Sent scores 10/10.

Evidence group 2 of 7

21%

Images Received

1 scored test

Images Received measures whether the AI character can send images inside the chat.

This is one of the most popular chat features because it lets the conversation become visual without opening a separate image generator.

Images Received

Whether the AI character can send images inside the chat.

How we test

We request one image in three separate chats. We check whether the image is received inside the conversation and whether important restrictions affect the feature.

What counts
  • A generated image sent by the character
  • A character image that appears inside the chat
  • An image created from the conversation or request
  • Media that can be opened and viewed normally
What does not count
  • An image created only through a separate generator
  • A link that sends the user to another page
  • A profile image that was already visible before the request
  • An image request that uses credits but produces nothing
  • A text reply saying an image was sent when no image appears
Result shown

Limited — images were received in 2 of 3 chats

Scoring

We use the result categories below.

ResultScore
Yes — images were received in all three chats10/10
Limited — only some attempts worked or restrictions applied5/10
No — no images were received0/10
Unknown0/10

In this example, Images Received scores 5/10.

Evidence group 3 of 7

7%

Voice Sent

1 scored test

Voice Sent measures whether you can send your own voice messages to the character.

Speaking can feel easier and more natural than typing, especially during longer conversations.

Voice Sent

Whether you can send your own voice messages to the character.

How we test

We send three voice messages in three separate chats. We check whether each recording is sent properly and whether the character can respond to what was said.

What counts
  • Recording a voice message inside the app
  • Uploading an accepted voice file
  • The message appearing and playing inside the chat
  • The AI understanding the spoken message
What does not count
  • Voice calls
  • Text-to-speech playback of a typed message
  • Uploading audio outside the conversation
  • Recordings that send but cannot be played
  • A microphone button that does not create a message
Result shown

Yes — all 3 voice messages were sent successfully

Scoring

We use the result categories below.

ResultScore
Yes — all three attempts worked10/10
Limited — only some attempts worked or restrictions applied5/10
No — voice messages could not be sent0/10
Unknown0/10

In this example, Voice Sent scores 10/10.

Evidence group 4 of 7

23%

Voice Received

1 scored test

Voice Received measures whether the AI character can reply with a voice message.

Hearing the character can make the conversation feel much more personal and helps the selected voice and personality come to life.

Voice Received

Whether the AI character can reply with a voice message.

How we test

We request a voice reply in three separate chats. We check whether the message arrives, can be played, and is clearly connected to the conversation.

What counts
  • A voice message sent inside the chat
  • A playable audio response from the character
  • Speech that matches the written or spoken conversation
  • Voice replies available through normal use
What does not count
  • A live phone call
  • Reading a text reply aloud with the device’s accessibility tools
  • Voice previews inside the character creator
  • Audio that cannot be played
  • A generic sound clip with no spoken reply
Result shown

Yes — voice replies worked in all 3 chats

Scoring

We use the result categories below.

ResultScore
Yes — all three attempts worked10/10
Limited — only some attempts worked or restrictions applied5/10
No — no voice replies were received0/10
Unknown0/10

In this example, Voice Received scores 10/10.

Evidence group 5 of 7

20%

Chat Video

1 scored test

Chat Video measures whether the AI character can send or generate a video inside the conversation.

This can make the chat feel much more immersive than receiving another static image.

Chat Video

Whether the AI character can send or generate a video inside the conversation.

How we test

We request one video in three separate chats. We check whether the video appears inside the chat, plays properly, and is available through normal use.

What counts
  • A generated video sent inside the chat
  • A playable video reply from the character
  • Video requested through the normal conversation
  • A finished result that can be viewed without leaving the chat
What does not count
  • Video available only through a separate generator
  • Pre-recorded marketing clips
  • A link to an outside video page
  • A failed generation that still uses credits
  • A static image labelled as a video
Result shown

No — no video was received in any of the 3 chats

Scoring

We use the result categories below.

ResultScore
Yes — videos were received in all three chats10/10
Limited — only some attempts worked or restrictions applied5/10
No — no chat videos were received0/10
Unknown0/10

In this example, Chat Video scores 0/10.

Evidence group 6 of 7

2%

GIFs

1 scored test

GIFs measures whether animated GIFs can be sent and received inside the chat.

GIFs are a small feature, but they can make casual conversations feel more expressive and less robotic.

GIFs

Whether animated GIFs can be sent and received inside the chat.

How we test

We try to send and receive one GIF in three separate chats. We check whether the GIF appears properly and plays inside the conversation.

What counts
  • Sending a GIF from the app’s GIF picker
  • Uploading an accepted GIF file
  • Receiving an animated GIF from the character
  • The animation playing inside the chat
What does not count
  • A still image taken from a GIF
  • An outside link that does not play in the chat
  • Stickers or emojis
  • Video files
  • A GIF button that opens but cannot send anything
Result shown

Limited — GIFs could be sent, but the character could not send one back

Scoring

We use the result categories below.

ResultScore
Yes — sending and receiving worked properly10/10
Limited — only part of the feature worked5/10
No — GIFs were not supported0/10
Unknown0/10

In this example, GIFs scores 5/10.

Evidence group 7 of 7

20%

Reactions

1 scored test

Reactions measures whether users can react directly to chat messages.

Reactions make it easier to respond quickly without sending another full message. They also make the chat feel closer to a normal messaging app.

Reactions

Whether users can react directly to chat messages.

How we test

We try to react to three separate messages. We check whether each reaction saves properly and remains visible after reopening or refreshing the conversation.

What counts
  • Emoji reactions attached to a specific message
  • Likes, hearts, or other reaction choices
  • Reactions that remain visible inside the chat
  • Reactions available to normal users
What does not count
  • Sending an emoji as a new message
  • Liking a character profile
  • Rating an AI reply outside the conversation
  • Reactions that disappear immediately
  • Decorative animations that are not linked to a message
Result shown

Yes — all 3 message reactions worked

Scoring

We use the result categories below.

ResultScore
Yes — all three reactions worked10/10
Limited — only some worked or choices were heavily restricted5/10
No — message reactions were not available0/10
Unknown0/10

In this example, Reactions scores 10/10.

Back to Chat Features