What Candy AI and HoneyChat offer
Chat and voice
Characters hold a dialog, “pick up your tone and remember what you told them”. Voice — separate voice messages and conversations with a choice of timbre: “soft and soothing, bold and confident”. Voice consumes tokens; how many tokens one message costs is not stated on public pages — not verified.
Photos and video
Image generation — “outfits, backgrounds, poses”; video — “watch them move and react”; separately — Live Action and Candy Cams, Shorts with new episodes every week. According to the terms, part of the media in character profiles is pre-made material created “using generative tools and licensed media modifications”.
Builder
“Create My AI Girlfriend”: ethnicity, age group, eye colour, hairstyle and hair colour, body, personality (from “chatty” to “intellectual”), voice, hobbies. The service warns: the more detailed the character, the more tokens generation consumes.
Memory
The service writes about memory a lot: “advanced personality modeling and memory retention”, “remembers your stories”. Not a single number: neither context size, nor storage period, nor number of facts — not verified. The only way to check is to tell the character something and come back a week later.
What is not disclosed
- The character's memory size — no numbers.
- The token cost of media and the price of extra tokens.
- The free account limits.
- Which AI models are used.
The case for Candy AI
For media Candy is the benchmark of the niche: video, Live Action and Shorts are rare. If the picture and motion are the main thing, it is the familiar place.
The case for HoneyChat
Media are included in the plan with clear limits: Basic — 10 images a day, 10 voice notes and 2 videos a month; Premium — 30 HD a day and 4 videos; up to 250 Ultra HD and 32 videos. No tokens. Memory is measurable and editable: from 8 facts about you on Basic to 60 on Ultimate, a story journal, 16K–64K context on the upper tiers.
