Leonardo.ai Problems & Issues

Free tier provides 150 tokens daily with no rollover. Since each image costs 5-8 tokens, you get roughly 18-30 images per day. Heavy users find this insufficient for meaningful engagement, forcing upgrades. Tokens reset at midnight UTC, not your local time.

Users report losing unused credits when canceling their subscription, calling the policy 'abusive.' Unlike some competitors that let you keep purchased credits, Leonardo.ai ties usage strictly to active subscription periods.

Users find the pricing very high, particularly impacting beginners and casual users. At $15-70/month for paid plans, it's expensive compared to alternatives, especially when free credits are insufficient.

Features like custom model training, higher token limits, and full ownership require higher-priced tiers. Users on lower plans find themselves needing to upgrade for features they assumed were included.

Video generation still needs improvement. Videos can get distorted as people move, with users reporting distortion when animating subjects like a child playing piano. Motion generation is not yet reliable for professional work.

Users still receive images with anatomical errors, including extra hands. One reviewer canceled their $10 monthly subscription specifically due to this issue. While improving with model releases, consistency issues remain.

Consistency issues may arise in character details across multiple poses. While slowly improving with new model releases, maintaining the same character appearance across different images remains challenging.

While Leonardo produces good results, users comparing directly to Midjourney note the artistic quality gap. Midjourney's default aesthetic is more polished for artistic work, though Leonardo excels at production-ready commercial imagery.

Users report occasional performance slowdowns, particularly during peak usage times. Generation queue times increase, and the interface may lag, affecting productivity for time-sensitive work.

Creating complex image elaborations isn't intuitive. Users find making images with specific styles or designs difficult through the interface. The learning curve for advanced features is steeper than expected.

Users note a learning curve around tokens and the feature set. Understanding token costs for different actions, model differences, and optimal settings takes time. Not as simple as Midjourney's type-and-generate approach.

Leonardo offers multiple models (Lucid Origin, Flux, Phoenix, custom) which provides flexibility but can overwhelm newcomers. Understanding which model to use for which purpose adds cognitive load.

Some users report being auto-subscribed and charged without clear notice, even when they didn't intend to continue. The subscription renewal catches users off guard, leading to unexpected charges.

In some cases, staff did not reply to account access issues, such as when users forgot their passwords. While general support receives praise, critical account issues may fall through the cracks.

All generations on the free plan are public, meaning others can see your prompts and images. Privacy requires upgrading to paid plans. This catches users off guard who expected private experimentation.