What is Leonardo AI? A guide to creating AI images with just a prompt

Ever typed and retyped the same prompt, hit generate a dozen times, and still couldn't get the result you wanted? The character's hair is black one time, brown the next; you finally get the camera angle right and the style drifts off somewhere else. With most AI image tools, using them feels like standing in front of a slot machine: you feed it a prompt and hope for the best, and whether it comes out good or bad is mostly out of your hands. That's fine if you just want one fun image, but it becomes a real headache when you need a consistent set of visuals for a brand, the same cast of characters across multiple scenes, or simply want to fix one detail without wrecking the whole picture. Leonardo AI exists to close that gap. Instead of leaving you at the mercy of the algorithm, it hands you back the controls: pick the right model, lock in a composition, fix a style, and edit directly on the image. In this article, let's dig into how to make the most of that control, so every image you generate relies a lot less on luck.

What is Leonardo AI?

Leonardo AI is an AI image generation and editing platform that runs right in your browser. You type a description in plain language (a prompt), and the system builds the corresponding image in seconds. So far, that sounds like every other image generator out there, but what sets Leonardo AI apart is that it isn't trying to be a machine that spits out random images. It's a workspace where you get to shape every step.

That strength comes from two things. First, Leonardo AI doesn't lock you into a single model. You can choose from several, including in-house models like Phoenix, Lucid Origin, and Lucid Realism, as well as integrated models like FLUX. Each model excels at a different kind of image, so the idea is to pick the right "engine" for each job instead of forcing one model to do everything. Second, wrapped around the image generation itself is a full set of steering and editing tools that let you nudge the output toward what you actually want, instead of starting over from scratch. Leonardo.Ai was founded in Sydney in December 2022 and was acquired by Canva in 2024, but it still runs as its own app and brand today.

Who is Leonardo AI best for?

Game artists and concept artists

This is the crowd that's been with Leonardo AI since its early days, when the tool was mostly known within the game development community. You can train a custom model to keep a character, item, or setting consistent across dozens of scenes, instead of getting a slightly different version every time you regenerate. That makes it far more realistic to build out a full, cohesive set of visual assets for a game project.

Marketers and brand content creators

For marketers, keeping visuals consistent matters a lot. Leonardo AI lets you lock in a style, color palette, and composition so an entire campaign looks like it was made by the same hand. You can also quickly sketch out ad concepts, product shots, or post illustrations without hiring a designer for every single piece.

Design freelancers and small studios

This group is usually juggling several projects at once on a tight budget. The on-image editing tools let you fix exactly the spot a client flagged without rebuilding the whole image, saving a ton of back-and-forth revision time. API access also makes it possible to plug image generation into an automated workflow.

People new to AI image generation

If you've never used an AI tool before, the free plan's 150 tokens a day is a comfortable way to try things out without spending anything. The web interface is intuitive, there's no software to install and no jargon to memorize; you type a description, see the result, and get familiar with the controls as you go.

Standout features

A diverse model library in one platform

Instead of locking you into a single fixed model, Leonardo AI lets you switch between many, from in-house models like Phoenix and Lucid to FLUX and dozens of community fine-tuned models. Each one is good at something different: some are built for photorealism, others specialize in illustration or anime-style characters. That means you pick the right tool for each task instead of forcing one model to carry the whole load.

An image steering toolkit

This is the feature set that really separates Leonardo AI from tools where you just type a description and hope for the best. Image Guidance lets you upload a reference image so the AI follows its composition, pose, or depth, while Elements lets you dial in how strongly a given style is applied. Instead of struggling to describe a complex composition in words, you hand the AI something visual to anchor to.

The Canvas editing workspace

Canvas is where you edit directly on the image: add detail to a specific area (inpainting), extend the frame beyond its original edges (outpainting), or stitch multiple pieces together. With Realtime Canvas, you can even sketch a few rough strokes by hand and watch the AI refine them into a finished image as you draw. This is what lets you fix exactly one spot without tearing down and rebuilding the whole picture.

Custom model training

If you need a character, product, or style to show up looking exactly the same across multiple images, you can upload a set of reference images and let Leonardo AI learn from them to build your own custom model. From there, every image you generate afterward follows what it learned. Most quick-turnaround image tools don't offer this, and it's a direct answer to the classic headache of "it comes out different every time."

A fairly complete production toolkit

Beyond static images, Leonardo AI also offers upscaling to boost resolution, short video generation from images, and 3D surface texture processing. By pulling multiple steps into one place, you can go from initial idea to finished asset without bouncing between half a dozen different programs.

Benefits of using Leonardo AI

Steer the result instead of regenerating endlessly

Because you can lock in composition and style from the start, results land closer to what you actually had in mind, and you spend less time re-prompting. That saves both time and tokens, instead of generating over and over until you get lucky.

Consistency for long-running projects

Thanks to custom model training and fixed styles, an entire brand's visual set, or a cast of characters across multiple scenes, can look cohesive. That's a clear advantage for work that needs systematic, structured image production rather than one-off, disconnected images.

Easy to get started, nothing to install

The whole platform runs in your browser, so there's no heavy software to install and no need for a high-spec machine. The interface is laid out intuitively enough that even first-time users can produce their first image within minutes.

A free plan you can actually put to the test

150 tokens a day, refreshing every 24 hours, lets you try out the core models and get a real feel for the quality before deciding whether to pay. It's a sensible way to understand how the token system works and pick the right plan, rather than upgrading too soon and realizing later it isn't the right fit.

Step-by-step guide to using Leonardo AI

Step 1: Create an account

Go to https://leonardo.ai/ and sign in with a Google, Canva, Apple, or Microsoft account to get started. Once you're in, you'll see the main workspace along with the feature menu in the sidebar.

Step 2: Open the image generation area and pick a model

Select Image Generation. In the left-hand column, open the Model selector: you can leave it on Auto and let the system pick the best model for your prompt, or open the Models list and search for a specific one. The list includes both models built by Leonardo and models integrated from third parties, so for beginners, starting on Auto and switching models later is the easiest approach.

Step 3: Write your prompt, adjust the settings, and generate

Type a description of the image you want, then adjust settings like size, aspect ratio, and number of images in the left-hand column. If you want the AI to follow a specific composition or reference image, you can add one via Image Guidance right here before generating. Once you're set, hit Generate, and it's worth starting with a handful of images to compare before spending tokens on a high-quality version.

Step 4: Edit the image in Canva

Once an image appears in your feed, hover over it, click to open it, and edit it in Canva to adjust it however you like.

Step 5: Upscale the image

Still hovering over the image, click "Upscale image" to boost the resolution for a sharper, clearer result. A settings panel will appear with sensible defaults; review them and click Upscale to begin.

Step 6: Download the image

Once you're happy with the result, click the download button directly on the image, or choose "Download Artwork" if you're working inside Canvas. On paid plans, exported images come without a watermark and can be used commercially.

Tips for getting the best results

Write structured prompts

Instead of typing a vague, general sentence, break your prompt into subject, setting, lighting, style, and camera angle. A description like "a woman standing by a window, warm afternoon light, cinematic style, eye-level angle" will give you far more consistent results than "a beautiful woman." The more specific you are, the less the AI has to guess.

Lean on steering tools instead of describing everything in words

For compositions or poses that are hard to put into words, upload a reference image through Image Guidance instead of trying to describe it in text. In our experience, the AI sticks far more closely to a visual anchor than it does to guesswork based on a vague prompt.

Watch your token usage and turn off features you don't need

Since every generation deducts tokens, keep an eye on which models eat through them fastest, especially heavier models and video generation. Turn off any options that are enabled by default if you don't need them, to avoid burning tokens unnecessarily, and consider training a custom model if you need the same subject repeatedly.

Leonardo AI pricing

Leonardo AI charges through a token system: every image you generate deducts a certain number of tokens depending on the model and settings used. The platform is split into three plan categories: Solo, Team, and API. For individual users, there are four main plans; prices below are monthly, with annual billing saving up to 20%.

• Free: $0, 150 tokens a day, refreshed every 24 hours. Images are generated publicly at basic quality settings. Good for experimenting and getting familiar with the core models.
• Essential: $12/month (about $9.60/month billed annually), 8,500 tokens/month with a token pool that can build up to 25,500. Includes private image generation, higher quality, unlimited saved collections, up to 10 custom trained models, and 2 concurrent jobs. This plan is also bundled into Canva Business. Good for individuals who use it regularly.
• Premium: $30/month (about $24/month billed annually), 25,000 tokens/month with a pool capped at 75,000, up to 20 custom trained models, 3 concurrent jobs, and unlimited standard-speed generation on select models. A good fit for freelancers and creators who use it often.
• Ultimate: $60/month (about $48/month billed annually), 60,000 tokens/month with a pool capped at 180,000, up to 50 custom trained models, 6 concurrent jobs, plus unlimited standard-speed image and video generation on select models. Built for professionals, small studios, and content producers.

• Team plans: start at $24/seat (Starter plan, $72/month total) and $48/seat (Growth plan, $144/month total), with a shared token pool, private generation for the whole team, and custom model training using shared tokens. Custom enterprise plans are also available.

• API plan: offered as pay-as-you-go, with $5 in credit to start, no commitment, and charges based only on actual usage; a good fit for teams building image generation into a product or automated workflow.

A note for readers in Vietnam: Leonardo AI is accessible normally through the browser, but the interface is currently English-only, and paying for a plan requires an international card (Visa, Mastercard). The pricing above was cross-checked directly against the official page at https://leonardo.ai/pricing/, and it's worth double-checking there before signing up, since prices can change over time.

Limitations of Leonardo AI

The token system runs out fast with heavy use

While the free plan sounds fairly generous, 150 tokens can disappear after just a few dozen generations, especially with heavier models or video generation. Regular users will likely need to manage their tokens carefully or move to a paid plan.

A learning curve for beginners

With so many models, tools, and settings, it's easy for newcomers to feel overwhelmed at first. Getting the most out of the steering tools takes some time to learn each control; it's not something you'll master right away.

Premium features sit behind higher-priced plans

Some genuinely valuable capabilities, like more generous video generation or a larger token allowance, are only available on the pricier plans. Users on lower tiers may start to feel boxed in as their needs grow.

Results still depend on how well you prompt

Like any AI image tool, if your description is too vague or contains conflicting instructions, the output can easily miss the mark, forcing you to try again. The real advantage here comes from how well you direct the tool, not from the underlying technology alone.

No Vietnamese interface yet

The entire platform is displayed in English, so anyone unfamiliar with the terminology may need a bit of time to adjust. Text rendering within images has improved but still isn't perfect, particularly with Vietnamese diacritics.

Overall

What makes Leonardo AI worth using isn't that it produces beautiful images; plenty of other tools can do that too. The real value is that it hands control back to you: choose your model, lock in a composition, fix a style, and edit exactly the detail that needs fixing. In other words, it's a creative studio you're in the driver's seat of, not a slot machine you leave to chance. If you only need the occasional fun image now and then, that level of control might be more than you need. But if you need consistent visuals for a brand, a long-running project, or day-to-day design work, that's exactly what makes the difference. The best way to get a feel for it is to start with the free 150-tokens-a-day plan, try steering a result toward exactly what you want, and see for yourself how different it feels to be the one driving instead of waiting on a random outcome.