AI companions are among the most monetisable consumer AI products. Users form ongoing relationships, return daily and — crucially — are willing to pay. But how they pay has changed.
1. The subscription
The original model: a free tier with limits, and a monthly or annual plan that unlocks longer memory, voice, images and unlimited messages. Subscriptions give predictable revenue and pair well with memory, which raises switching costs.
Trade-off: heavy users can cost far more in inference than they pay, while light users churn.
2. Tokens and credits
Increasingly common, especially for image, voice and video generation. Users buy packs of tokens and spend them on premium actions. Tokens align revenue with compute cost and let the most engaged users spend far more than a subscription cap would allow.
Trade-off: pricing complexity, and scrutiny from app stores and regulators over spending by vulnerable users.
3. Hybrid models
Many leading platforms now combine a subscription with token top-ups — the subscription covers everyday chat, tokens cover expensive media.
4. Creator and fan models
A newer pattern borrows from the creator economy: companions built around a persona or niche, monetised through memberships, tips, photo unlocks and custom requests. This is where white-label infrastructure has grown fastest, letting affiliates and media buyers launch branded companion sites and share revenue with the platform that runs them.
5. B2B and licensing
Some companies sell companion technology to other businesses — for wellbeing, education, elder care or customer engagement — rather than to consumers directly.
What's next
Expect more experimentation with usage-based pricing as video avatars and real-time voice push compute costs up, and more pressure from regulators around spending protections. The winners will be those who price in a way users find fair and that covers the true cost of every conversation.
