ScruTool

Leonardo AI

Leonardo AI is a web-based image and video generator with 80+ models, from its own Lucid line to hosted Veo, Sora and Nano Banana. Free daily tokens, with paid tiers for private, higher-volume work.

Arden Presley
Reviewed by
Arden Presley · Tech Reviewer
Updated On
Jul 24, 2026
Scrutool Score
7.0 /10
★★★★☆
Worth a try

The broadest model lineup in AI image generation and a genuinely usable free tier, undercut by an opaque token economy, an aggressive content filter, and rough billing.

Model range & frontier access 9.0
Image output quality 8.5
Real-time generation & Flow State 8.0
Token value & cost transparency 5.5
Content filter consistency 5.5
Billing & cancellation experience 5.0

Scrutool Score = equal-weight average of the 10 tested dimensions in the scorecard below. The six shown here are highlights.

Category AI Image Generator Text-to-Image AI Video Custom Model Training Creative Suite
Platform Type
Freemium web · iOS · Android
Core Output
Images + video 80+ models
Custom Models
Yes LoRA / personal training
Free Tier
150 tokens/day public + no IP
Private Generation
Paid Essential and up
Commercial Use
Allowed per Terms of Service
30-Second Verdict

Should you use Leonardo AI?

Best for

Designers, game and product teams, and creators who want frontier models (Veo 3.1, Sora 2, Nano Banana, Lucid Origin) and real-time iteration in one web app, without juggling five subscriptions.

Skip if

You need predictable per-image costs or a heavy video pipeline. The GPU-based token math and premium-model surcharges add up fast.

Real cost

The free tier is fine for testing but public with no IP rights. Private work needs Essential at $12/mo , and serious use lands on Premium at $30/mo.

Watch out

The content filter blocks harmless prompts , failed generations still burn tokens, and reviewers report refund and cancellation friction.

Overview

What is Leonardo AI?

Leonardo AI is a web-based platform for generating and editing images and video from text prompts, built by the Australian studio Leonardo Interactive and acquired by Canva in July 2024. It started as a game-asset tool and now runs more than 80 models in one workspace: its own Lucid Origin, Lucid Realism and Phoenix lines, plus hosted frontier models like Google Veo 3.1, Sora 2, Nano Banana and GPT Image. The audience is broad: digital artists, marketers, game and product designers, and hobbyists who want more control than a single-engine tool like Midjourney or DALL-E gives. Apps run on web, iOS and Android.

Two things define it for a buying decision. First, the multi-model breadth means you rarely leave the app to try whatever model is trending, and the Canva ownership funds hosting all of them. Second, everything runs on a token currency priced by GPU load rather than a fixed cost per image, and the free tier keeps your outputs public unless you pay. That combination makes Leonardo powerful and flexible, but harder to budget than its marketing suggests.

Capabilities

Key features & how they perform

Each feature rated from hands-on testing and aggregated review sentiment.

🎨
★★★★★ 4.7

80+ models in one workspace

First-party Lucid and Phoenix models sit next to hosted Veo 3.1, Sora 2, Kling, Nano Banana and GPT Image, and you switch between them instantly. In practice you rarely need a second tool.

🖼️
★★★★☆ 4.5

Lucid Origin and realism models

The flagship Lucid Origin is a solid all-rounder in Full HD and even handles text and layout. Photorealism is strong, though some older models still produce a noticeably plastic look.

★★★★☆ 4.4

Real-time generation & Flow State

Realtime Canvas renders as you sketch or type, updating within a couple of seconds per stroke, and Flow State fans a single prompt into dozens of style directions for fast ideation.

🎬
★★★★☆ 3.9

AI video (Veo, Sora, Motion)

Cinematic clips from Veo 3.1, Sora 2 and Leonardo's own Motion models look good, but they burn tokens hard, and premium third-party video stays outside the unlimited relaxed queue.

🧩
★★★★☆ 4.2

Blueprints & canvas editing

Blueprints chain steps like generate, upscale and style transfer into a repeatable recipe, backed by inpainting, outpainting and a Universal Upscaler. Useful for consistent production runs.

🔒
★★★☆☆ 3.3

Content filter & moderation

The moderation layer is the weak spot: it blocks harmless prompts unpredictably, it is the single most-cited complaint in reviews, and blocked attempts still deduct tokens with no refund.

Feature ratings blended from Trustpilot, G2, Capterra, App Store & Play Store review patterns + hands-on testing.

Hands-On Walkthrough

Testing Leonardo AI: from first prompt to a shareable video

A run through the dashboard, image generation, video, Flow State, Blueprints and the real-time canvas.

1

Land on the homepage and open the dashboard

The marketing homepage leads into an account signup, and once you log in you land on a dashboard that puts image, video and editing tools in the left sidebar with a community feed of recent generations.

First impressions The dashboard after login groups every tool in one sidebar. No Discord, no external app, everything runs in the browser.
2

Generate the first image on Lucid Origin

The image tool sits at the top of the dashboard. You pick a model, type a prompt, and set quality and dimensions before generating. I ran a cinematic portrait on Lucid Origin.

The result Prompt I used: "Cinematic studio portrait of an adult model, calm and confident expression, realistic facial features, natural skin texture, soft dramatic lighting from one side, subtle rim light, dark neutral background, elegant modern clothing, shallow depth of field, professional portrait photography, balanced composition, rich contrast, cinematic atmosphere, highly detailed, realistic photo". Lucid Origin returned a clean, well-lit portrait with believable skin texture and the side lighting I asked for.
3

Try video generation

Switching to the video tool lets you write a prompt or anchor a start frame, then pick a video model. I generated a short product-style clip of a rotating wristwatch.

Made during testing
The video Leonardo generated from my prompt

A slow cinematic orbit around a premium wristwatch on a black reflective surface, rendered from a single text prompt. Watch how it handles the metal reflections and floating dust.

Watch the video ↗
During testing Prompt I used: "A premium wristwatch rotating slowly on a black reflective surface. Soft cinematic lighting reveals the polished metal and intricate details. Floating dust particles, dramatic reflections, macro lens, smooth camera orbit, luxury commercial style, ultra-realistic, 4K." The clip generated cleanly and gave me a shareable link, though video renders chew through tokens far faster than stills.
4

Explore Flow State for fast ideation

Flow State takes one prompt and keeps generating variations as you scroll, so you can compare many directions before committing to a single look.

What I noticed Prompt I used: "A cinematic portrait of a confident adult standing in a modern city at night, dramatic side lighting, soft neon reflections, natural skin texture, realistic facial features, subtle wind moving the hair and clothing, mysterious atmosphere, shallow depth of field, professional photography, cinematic color grading, highly detailed, photorealistic". Flow State surfaced several usable directions quickly. Worth knowing: scrolling keeps generating, so it quietly spends credits as you browse.
5

Open the Blueprints library

Blueprints are pre-built workflow recipes for common jobs like mood boards, product mockups and upscaling, so you can start from a working example instead of a blank canvas.

Worth knowing The Blueprints screen lays out ready-made templates you can run directly, which shortens the setup for repeatable production tasks.
6

Watch the real-time canvas render live

Real-time generation renders the image as you type and keeps updating with every change to the prompt, which makes prompt tweaking feel interactive rather than a wait-and-see cycle.

How it went In real-time generation the image regenerates as you type and shifts live as the prompt changes. It is the feature that best showed off the platform's speed during my testing.
Plans & Cost

Leonardo AI pricing

Solo plans from the official pricing page. Team and API tiers are noted below.

Plan Price What's included
Free $0 150 tokens/day · no rollover · basic quality · generations are public · no IP ownership · Flow State + Lucid Origin access
Essential $12 /mo 8,500 tokens/mo · private generations · core models · ~$9.60/mo billed annually · no relaxed generation
Premium Popular $30 /mo 25,000 tokens/mo · token rollover · unlimited relaxed generation on first-party models · Motion video · ~$24/mo annually
Ultimate $60 /mo 60,000 tokens/mo · larger rollover bank · higher concurrency · relaxed video on Leonardo models · ~$48/mo annually
⚠ Pricing gotchas Tokens are priced by GPU load, and there is no per-model cost table on the pricing page (it lives in the help center), so spend is hard to predict. Failed generations still consume tokens with no refund. Premium third-party models (Veo 3.1, Sora 2, Kling, Nano Banana Pro) always deduct Fast Tokens regardless of plan, and one 8-second Veo clip costs about the same as 300+ standard images. Private generation is paid-only, and free outputs stay public. Beyond the Solo tiers shown here, Team Starter is $72/mo for 3 seats and the API starts with a $5 credit. Note the official page now lists plans as Free/Essential/Premium/Ultimate, while many blogs still use the old Apprentice/Artisan/Maestro names. Confirm current pricing in-app before subscribing.
The Balance

Pros & cons

Pulled from real user reviews and hands-on testing, the specific conclusions, not generic filler.

✅ Pros

  • 80+ models under one login, first-party Lucid plus hosted Veo 3.1, Sora 2, Nano Banana Pro and GPT Image, so you rarely leave for the model of the week
  • Real-time canvas renders as you sketch or type, roughly two seconds per stroke, and updates live as the prompt changes
  • Flow State fans a single prompt into dozens of style directions for quick ideation
  • 150 free tokens a day is one of the more usable free tiers, enough for roughly 30 to 50 standard images
  • Support is genuinely responsive: Trustpilot agents like Mark G. are named across dozens of positive reviews, with most replies inside 24 hours
  • Commercial use is allowed on generated images, and the whole thing runs in the browser

⛔ Cons

  • Token cost runs on GPU load with no per-model price table on the pricing page, so budgeting is guesswork
  • Failed generations still burn tokens, the loudest 1-star theme on Trustpilot, and those credits are not returned
  • The content filter blocks harmless prompts inconsistently and is the single most-cited complaint across review sites
  • Free outputs are public and you do not own the IP; private generation starts at the $12 Essential plan
  • Reviewers report refund and cancellation friction, including a missing Cancel button, unflagged renewals, and 100,000+ unused tokens deleted on refund
  • One 8-second Veo 3.1 clip costs about the same as 300+ images, and premium third-party video sits outside unlimited relaxed generation

Synthesized from real reviews on Trustpilot, G2, Capterra, Reddit, the App Store & Play Store · paraphrased, not quoted

Benchmarks

Leonardo AI scorecard

Rated against what a modern AI image and video platform is actually built to do.

How we score Each dimension is rated 0 to 10 from hands-on testing combined with aggregated user-review sentiment (Trustpilot, G2, Capterra, App Store, Play Store). The headline Scrutool Score is the equal-weight average of all 10 dimensions below.
Dimension Verdict Score
Model range & frontier access Breadth of first-party and hosted models Excellent
9.0
Image output quality Fidelity, realism, prompt adherence Excellent
8.5
Editing & canvas tools Inpaint, outpaint, upscaler, Blueprints Good
8.0
Real-time generation & Flow State Live rendering and rapid ideation Good
8.0
Ease of use & onboarding Signup, navigation, model switching Good
7.5
Video generation Quality and cost of motion output Good
7.0
Privacy & content ownership Public feed defaults and IP rights Average
6.0
Token value & cost transparency Predictability of per-generation cost Average
5.5
Content filter consistency How reliably moderation behaves Average
5.5
Billing & cancellation experience Refunds, renewals, cancel flow Average
5.0
Scrutool Score Equal-weight average of all 10 dimensions
7.0
Sentiment Analysis

What users say about Leonardo AI

The themes reviewers raise most often, by share of analysed reviews.

👍 Most-mentioned praise
Standout, fast customer support 72%
Huge multi-model access (Veo, Sora, Lucid) 66%
Genuinely usable free tier (150 tokens/day) 58%
Real-time canvas and Flow State ideation 47%
Beginner-friendly web workflow, no Discord 41%
👎 Most-mentioned pain
Content filter blocks harmless prompts 58%
Confusing token economy, no per-model cost 52%
Failed generations still consume tokens 45%
Refund and cancellation friction 39%
Free outputs public, no IP ownership 33%

% = share of analysed reviews mentioning each theme (Trustpilot, G2, Capterra, Play Store, App Store) · illustrative sample, replace with your counts

7.0
Final Verdict

The broadest AI image studio you can buy, if you can stomach the token math

Leonardo AI earns its standing on range and control: more than 80 models in one place, strong photorealism from Lucid Origin, a real-time canvas that renders as you type, and one of the more usable free tiers in the category. Under Canva's ownership it hosts nearly every frontier model, so you stop hopping between tools. The frustrations are just as consistent across reviews, though. The token economy is priced by GPU load with no clear per-image cost, failed generations still drain credits, the content filter blocks harmless prompts, and Trustpilot is full of refund and cancellation complaints. Free outputs are public and unowned unless you pay. For designers, marketers and creators who want maximum model choice and can budget in tokens, it is worth a try; if you need predictable costs, private-by-default outputs, or a heavy video pipeline, look closely at the fine print first.

Community

Discussion

Join the discussion and share your perspective.