MagicAIMagicAI
All ideas

πŸ”₯A Smarter Way to Handle AI Credits (That Users Will Love)πŸš€

Hello, I would like to suggest an improvement to the subscription system, or ask if something like this is already possible. Currently, credits/tokens are assigned separately to each model inside a package. While this works, it creates a problem in real usage. Most users do not use all models β€” they usually focus on only 2–3 models. Example: A user buys the β€œUltimate” plan for $59, which includes multiple video models (for example Kling models), each with a limit like 5 videos per model. However, the user mainly uses only one model (for example: kling-video/v2.6/standard/motion-control). In this case, the limits for other models are unused and wasted, which makes the subscription feel inefficient and not worth the price. This may result in users not renewing their subscription. Possible solution (suggestion): Instead of assigning limits per model, introduce a shared credit pool per category. Example: Ultimate plan includes: 1500 video credits All active video models use the same pool Each model has a different cost depending on complexity, for example: Sora 2 Pro β†’ 300 credits Kling v2.6 Motion Control β†’ 120 credits Grok Image-to-Video β†’ 180 credits So no matter which model the user selects, credits are deducted from one shared pool. Benefits: More flexibility for users Users pay for what they actually use Better user experience Higher chance of subscription renewal Easier package management (no need to configure limits per model) Additional idea: It would also be very helpful to show the credit cost per model in the UI, so users know how much each action will consume before generating. This example is based on video models, but the same approach could also be applied to chat models and image generation. Of course, the credit cost for each model can be defined individually, since every model has a different cost on the backend (some are cheaper, some are more expensive). I believe this approach would make the system much more flexible, scalable, and user-friendly. Thank you!