How to Use Leonardo AI (Image Generation Guide)

Learning how to use Leonardo AI opens up one of the most versatile image generation platforms available today. Where Midjourney excels at artistic quality and DALL-E excels at accessibility, Leonardo sits in the middle: strong image quality, a proper web interface, fine-tuned models for different styles, and a canvas editor for post-generation work. It also has a generous free tier, making it a solid place to start if you're new to AI image generation.

If you've tried creating images with ChatGPT and want more control over the output, Leonardo is a natural next step.

Getting Started

Step 1Create Your Account

Go to leonardo.ai and sign up with Google, Apple, or email. Free accounts receive 150 fast tokens per day, which reset every 24 hours. That's typically enough for 15-30 image generations daily, depending on the model and settings you choose.

Step 2Generate Your First Image

Click "Create" from the dashboard. You'll see a prompt box and a panel of settings on the left. Type a description of the image you want, select a model (start with Phoenix, more on this below), and click "Generate."

Your first image will appear in about 10-30 seconds on a paid plan, or potentially longer during peak hours on the free tier.

If you want to build strong AI image generation skills from the ground up, the AI Academy covers the fundamentals across multiple tools, including prompting, model selection, and creative workflows.

Step 3Understand the Token System

Leonardo uses a token-based system rather than a fixed number of generations. Different actions cost different amounts of tokens:

• Basic image generation: ~8-12 tokens per image
• Higher resolution or more complex models: 15-24 tokens
• Image-to-image transformations: varies by complexity
• Video generation: significantly more tokens

Paid plans offer more tokens, faster generation, and access to premium features.

Choosing the Right Leonardo AI Model

Leonardo's biggest differentiator is its variety of models, each tuned for different use cases. Here are the ones that matter:

Phoenix (Recommended Default)

Phoenix 1.0 is Leonardo's flagship model and the best starting point for most users. It delivers 95% prompt adherence (compared to 70-80% for standard models), which means it actually generates what you describe. Phoenix also renders legible text in images, something most AI image models still struggle with. If your prompt includes a sign, label, or title, Phoenix will often get the text right.

Best for: Marketing assets, social media graphics, product mockups, any image where you need the output to match your description precisely.

Leonardo Diffusion XL

A versatile model based on Stable Diffusion XL. It produces high-quality photorealistic and artistic images with good consistency.

Best for: General-purpose image generation, photography-style outputs, concept art.

Leonardo Anime XL

Specialized for anime and manga-style artwork. If you're creating content in these styles, this model produces significantly better results than generic models.

Community Models

Leonardo hosts user-created fine-tuned models trained on specific styles, characters, and aesthetics. Browse the model library for specialized options; there are models fine-tuned for everything from pixel art to architectural visualization.

Leonardo AI Prompting Tips for Better Results

Leonardo's Phoenix model responds well to structured, specific prompts. Here's how to write them effectively.

Start Small and Build

Begin with a short, clear description: "A golden retriever sitting in a sunlit meadow." Generate, assess the result, then add details in subsequent generations: "A golden retriever sitting in a sunlit meadow, wildflowers in the foreground, soft bokeh background, warm afternoon light, photograph style."

Resist the urge to write a 200-word prompt on your first try. Shorter prompts give you a clearer understanding of what each element contributes.

Use Negative Prompts

Leonardo's negative prompt field is powerful. Add things you explicitly don't want: "blurry, distorted hands, low quality, watermark, extra limbs, text artifacts." This significantly reduces common AI image issues.

Specify Style Explicitly

Don't leave style to chance. State it directly:

• "Digital illustration style"
• "35mm film photography"
• "Flat vector design, clean lines"
• "Oil painting with visible brushstrokes"
• "Studio Ghibli-style animation"

Handle Text Carefully

When you need text in your image (and you're using Phoenix), enclose the exact words in double quotation marks within your prompt: A coffee shop sign that reads "Morning Brew". Keep text strings short; single words or short phrases work best.

Advanced Leonardo AI Features

AI Canvas

Leonardo's Canvas is an interactive editor that goes beyond basic generation. You can:

• Inpaint: Select a region of an existing image and regenerate just that area. Fix a face, change an object, or modify a background without regenerating the entire image.
• Outpaint: Extend the borders of an image. Start with a portrait and expand it into a landscape by generating the surrounding environment.
• Sketch-to-image: Draw a rough sketch, and Leonardo generates a polished version following your composition.

Canvas is where Leonardo really separates itself from simpler text-to-image tools. If you're using AI-generated images as part of a broader content creation workflow, the ability to refine outputs in-platform saves significant time.

Developing a complete creative pipeline with tools like Leonardo is the kind of practical skill our AI Academy is designed to teach, with guided projects and real use cases.

Image-to-Image

Upload an existing image and use it as a reference for generation. Control the "strength" parameter to determine how closely the output follows the original: low strength for close resemblance, high strength for loose inspiration. This is useful for creating variations, changing styles, or using reference photos as starting points.

Real-Time Generation

Leonardo's real-time canvas lets you see results update as you type and modify your prompt. It's fast and useful for iterating quickly on compositions and styles.

Video Generation

Leonardo now includes video generation capabilities, allowing you to animate still images or generate short clips from text prompts. It supports its own Motion model as well as integration with Google's Veo 3 for higher quality output. For more advanced video work, see our guides on Sora AI and Runway AI.

Leonardo AI Pricing

The free tier is genuinely useful for learning and casual use. The main limitations are slower generation during peak hours (8-20 minute waits are common) and restricted access to some premium models and features.

Starter removes queue wait times and unlocks more tokens. If you're generating images regularly, this is the minimum recommended plan.

Creator adds priority servers, commercial usage rights, and access to all premium features. This is the sweet spot for professionals and freelancers.

Leonardo vs. Midjourney vs. DALL-E

Leonardo wins on control and versatility. Multiple models, Canvas editing, image-to-image, and the best free tier in the category. If you want to fine-tune your outputs and do post-generation editing, Leonardo is the most capable platform.

Midjourney produces the highest-quality artistic images with less effort. Its aesthetic sense is unmatched, and the default output simply looks better. The tradeoff: it runs through Discord (less convenient), offers fewer editing tools, and the free tier is minimal.

DALL-E 3 (via ChatGPT) is the most accessible. No separate account needed, natural language prompts work well, and it's included with ChatGPT Plus. Quality is good but you have less control over models, styles, and post-generation editing.

For most beginners, the recommendation is: start with DALL-E in ChatGPT for convenience, move to Leonardo when you want more control, and add Midjourney when you need premium artistic quality.

Knowing which tool to use for which job is half the battle. The AI Academy helps you build that judgment with structured comparisons and hands-on practice across the leading platforms.

Practical Tips

• Save prompts that work. When you get a great result, copy the prompt to a notes file. You'll build a personal library of reliable techniques.
• Use the community. Leonardo's community gallery shows other users' images alongside their prompts. This is one of the fastest ways to learn what works.
• Batch similar requests. If you need five images in the same style, generate them in the same session using the same model and similar prompts. This produces more consistent results than generating one at a time across different sessions.
• Check commercial rights. Free-tier images have restricted commercial use. If you're generating images for business purposes, make sure your plan includes commercial licensing.

FAQ

Want to go beyond image generation and build a complete AI toolkit? Start your free 14-day trial →