Leonardo.ai

AI image and video generator with advanced creative control.

Price: Free plan with 150 daily tokens. Paid plans from $12/month (Apprentice): $12/month, Artisan $30/month, Maestro $60/month. Up to 20% off with annual billing. API from $9/month.

Platforms: Web · iOS · Android

Free plan

API available

Open source

Available in Spanish

No watermark

What is Leonardo.ai?

Leonardo.ai is an AI-powered image and video generation platform offering granular control over creative parameters. Built for designers, game developers and content creators, it stands out for letting you train custom models on your own images, keep brand visuals consistent and edit in real time with AI Canvas. It includes video generation with models such as Google's Veo 3, preset blueprints and guaranteed commercial rights even on the free plan.

Pros

Very generous free plan (150 tokens/day) at professional quality, ahead of Midjourney and DALL-E
Trainable custom models for guaranteed brand consistency
Real-Time Canvas for advanced editing (inpainting, outpainting) on the fly
Built-in video generation with Veo 3 and Motion, plus image animation
Robust API with several enterprise integration options

Cons

English interface (it accepts Spanish prompts, but results are better in English)
The free plan watermarks images and grants no commercial rights
Steep learning curve given the sheer number of options and settings
Daily tokens do not roll over: they are lost if unused within 24 hours

Frequently asked questions

Can I use images generated on the free plan commercially?

No. Commercial rights are only available on paid plans (from $12/month). The free plan is for personal use only.

Does Leonardo.ai work in Spanish?

The interface is in English, but it understands Spanish prompts perfectly. That said, results are usually better with English prompts because the models were trained mostly on English data.

How many images can I generate with 150 free tokens?

Between 18 and 30 images a day at standard resolution (512x512), depending on the model and resolution chosen. Fast models use fewer tokens.