Hands-on benchmark · Tested by Rajat Chauhan · August 2026

Leonardo AI review: nine prompts tested, first output kept and scored

Item: Leonardo AI
Rating: 8.3
Author: Rajat Chauhan

A photoreal specialist that also draws the best anime we tested; weakest on hands

Leonardo has many models, so we locked it to one, Lucid Origin, its own flagship, with automatic selection off, and ran the same nine prompts as every other tool: default settings, first output kept. Every output is below.

See all nine outputs Read the verdict Visit Leonardo AI ↗

Leonardo · at a glance

MakerLeonardo.Ai (Canva)

Model testedLucid Origin (flagship)

Free tierYes, daily tokens

Paid from~$12/mo (Apprentice)

Best atAnime and illustration (the best in our benchmark), plus elite photoreal and product shots.

Weak atPrecise instructions: it follows the mood of a prompt over its exact details, like noon rendered as golden hour.

Bad atHands. It failed the exchange prompt outright, and its fingers do not survive close inspection.

9identical prompts, every tool

1stoutput kept, no re-rolls

3axes scored per image

0sponsored placements

The nine tests, every output

Each prompt was pasted verbatim into Leonardo on the Lucid Origin model; the image below is the first one it returned. Scores are Adherence, Coherence, and Aesthetic on a 0 to 10 scale, with the reasoning under each.

P01

TEXT RENDERING

a neon sign above a ramen shop that reads "OPEN UNTIL 3AM", night street, shallow depth of field

7Adherence

7Coherence

9Aesthetic

7.7Image avg

✓“OPEN UNTIL 3AM” is spelled correctly across three lines of red neon
✓Its top aesthetic mark: red tube-glow spilling down a tiled facade, the sign the only light in frame
✗Shot flat and frontal against a wall, so the shallow depth of field clause never gets a chance
✗The awning fascia degrades into pseudo-text, reading “RamEar INTGIL” and “Amtint Loam”
•Solid rather than standout on text, the letters hold but the scene does the heavy lifting

P02

TEXT RENDERING

a product box for "NOVA MINT" toothpaste, brand name clearly printed, teal and white packaging, studio light

7Adherence

6Coherence

8Aesthetic

7.0Image avg

✓The hero brand name is roughly right on the box
✓Packaging shape and studio presentation are broadly correct
✗The taglines break down into unreadable text
✗Weak depth of field flattens the packaging shot
•The clean brand word versus broken body text is the usual text split, its lowest coherence

P03

PHOTOREALISM

close-up portrait of a 60-year-old fisherman, weathered skin, natural window light, 85mm lens, no retouching

9Adherence

9Coherence

10Aesthetic

9.3Image avg

✓An elite fisherman portrait with accurate age and real skin texture
✓Superb natural light, honouring the "no retouching" clause
✓An unedited, documentary feel rather than a smoothed, glamour look
✓Sound facial anatomy with no artefacts on close inspection
✓Its top aesthetic mark, and among its best images overall

P04

PHOTOREALISM

street photo of a rainy Tokyo crossing at night, wet asphalt reflections, neon signage, 35mm documentary style

9Adherence

8Coherence

9Aesthetic

8.7Image avg

✓The only image in the whole test to render actual falling rain streaks
✓Warm lamp-light pools across the wet zebra stripes, a manhole cover catching the reflection
✓One umbrella-carrying pedestrian mid-crossing, a single car’s headlights behind: a true 35mm documentary read
✗Vertical shop signage reads as plausible kana at a glance, but the characters form no real words
•A genuinely distinctive take, the rain sets it apart from every rival

P05

HANDS & ANATOMY

two hands exchanging a coffee cup, close-up, all ten fingers visible, natural pose

5Adherence

6Coherence

8Aesthetic

6.3Image avg

✗Its clear weakness: it failed the exchange, rendering one person's static two-handed hold, not a handoff
✗The front hand's fingers are misproportioned and merge on close inspection
✗The depicted action is wrong, missing the "exchanging" verb entirely
•Its lowest score, and the lowest hand score in the entire benchmark
•Aesthetic still holds up, the image looks fine until you check the hands

P06

PRODUCT SHOT

a skincare serum bottle with a dropper, minimalist beige background, soft reflections, premium cosmetics photography

9Adherence

9Coherence

9Aesthetic

9.0Image avg

✓An ad-grade serum shot with clean glass and a correct dropper
✓A believable premium studio look throughout
✓Correct proportions and materials on the bottle
✓Even, balanced sub-scores across adherence, coherence, and aesthetic

P07

ANIME & STYLE

anime-style portrait of a young woman with short blue hair, school uniform, soft cel shading, detailed eyes

9Adherence

10Coherence

9Aesthetic

9.3Image avg

✓The best anime in the entire benchmark: genuine soft cel shading
✓An authentic sailor uniform, exactly the wardrobe clause requested
✓Detailed eyes and a true drawing rather than a photo in disguise
✓The only tool in the test to fully nail this prompt
✓Its top coherence mark, a clean, consistent illustration

P08

PROMPT ADHERENCE

a red cube on top of a blue sphere, to the left a green pyramid, plain white studio background, even lighting

10Adherence

9Coherence

8Aesthetic

9.0Image avg

✓Passed the shapes prompt cleanly: a real green pyramid correctly placed
✓The cube sits on the sphere exactly as asked, with correct colours
✓A genuine square-based pyramid, not a tetrahedron substitute
✓Its top adherence mark, a full spatial pass
✗The balance is a touch too clean, costing coherence and aesthetic

P09

PROMPT ADHERENCE

a split scene: left half desert at noon, right half snowy forest at night, clear vertical divider down the middle

7Adherence

9Coherence

9Aesthetic

8.3Image avg

✓A beautiful split with clean contrast and a tidy central divider
✓High coherence and aesthetic, a gorgeous render in isolation
✗Rendered golden hour instead of the specified desert-at-noon
•A clear case of following mood over the precise instruction

THE VERDICT

Leonardo, on its Lucid Origin model, is a specialist. It produced elite photorealism, ad-grade product shots, and the single best anime illustration in our entire benchmark, the only tool to nail genuine soft cel shading. Its weaknesses are just as specific: it failed the hand-exchange prompt outright, and it rendered our split scene at golden hour when the prompt asked for noon. A brilliant tool that rewards knowing what it is good at.

Best for

Anime and illustration, photoreal portraits and scenes, and clean product shots, especially for creators who want fine control over models and styles.

Watch for

Weak hand anatomy, garbled text beyond a headline word, and results that follow the mood of a prompt more faithfully than its precise details.

⚠ Which Leonardo we tested

Leonardo offers several image models and an automatic mode that picks one for you. Automatic can route the same prompt to very different engines, including third-party models, which makes a mixed set impossible to score as one tool. So we turned automatic off and ran all nine prompts on Lucid Origin, Leonardo's own flagship model, for a fair, single-engine result.

This matters: on automatic, some prompts route to models like GPT Image, which behave differently (in our spot checks, they rendered legible foreign-language signage that Lucid Origin garbles). The score below reflects Lucid Origin specifically, not Leonardo's automatic mode.

Category scores, broken down

Two prompts per category (one each for hands and product), averaged. The overall 8.3 is the mean of all nine images. Note the spread: elite at the top, genuinely weak at the bottom.

Anime & style9.3

The best in our whole benchmark. Genuine cel shading, authentic uniform.

Photorealism9.0

Elite faces, and the only tool to render actual falling rain.

Product shot9.0

Ad-grade glass, liquid, and a correctly straight dropper.

Prompt adherence8.7

Aces the shapes prompt; misses noon for golden hour on the split scene.

Text rendering7.3

Hero words correct; taglines garble and depth of field is weak.

Hands & anatomy6.3

Its weak spot. It rendered a static hold, not an exchange.

Strengths, and the specific weaknesses

Every point below traces to a specific image above. Leonardo has the widest spread of any tool we've tested, which makes naming the trade-offs essential.

Strengths

Where Leonardo excels

The best anime in our benchmark. Its 9.3 illustration was the only one to nail genuine soft cel shading, an authentic sailor uniform, and detailed eyes as a true anime drawing rather than a photo. No other tool matched it on this.
Elite photorealism. The fisherman portrait reads the age accurately and honors the no-retouching clause with real skin texture, and the Tokyo scene is the only image in our whole test to render actual falling rain.
Ad-grade product shots. The serum bottle scored 9.0, with convincing frosted glass, a correctly straight dropper, and a drop of serum suspended mid-fall.
Clean spatial logic when it counts. It passed the hard shapes prompt completely: a real green pyramid, correctly placed, with the cube on the sphere exactly as asked.
Genuine model control. Locking to a single flagship model gave a consistent, high-quality result. Creators who want to choose and tune models have real depth here.

Trade-offs

Where it falls short

Hands are its clear weakness. It failed the exchange prompt, rendering one person's static two-handed hold instead of a handoff, and the front hand's fingers are misproportioned and merge on close inspection. Its lowest score.
It follows mood over precise instructions. Asked for a desert at noon, it produced a beautiful golden-hour scene instead. The lighting is gorgeous and the instruction is wrong, which is exactly the trade-off to know about.
Text garbles beyond a headline word. It rendered a neon sign correctly but turned packaging taglines into nonsense, and its foreign-language signage garbles where some other models render it cleanly. For blocks of legible text, look elsewhere.
Depth of field can miss. On the neon prompt the requested shallow focus came back mostly deep. Small, but part of the pattern of following the scene more than the spec.

Our testing method, and why it's fair

Most best-of lists rank tools the author never opened. We do the opposite, and the method is deliberately strict so the scores mean something.

Every tool gets the identical nine prompts, pasted verbatim. We use default settings and keep the very first image, with no re-rolls and no picking the best of a batch. For Leonardo, that meant one decision up front: because its automatic mode mixes different models, we locked it to a single flagship (Lucid Origin) so the nine scores measure one consistent engine, comparable to every other tool.

We sell nothing on this list, so there is no commercial reason to rank one tool over another. The scores are the scores.

Same nine prompts, every tool

Two prompts each for text, photorealism, and spatial adherence, plus one each for hands, product, and anime.

One model, first output

Lucid Origin with automatic selection off, keeping the first image, no re-rolls, so the result reflects a single engine.

Three axes, full resolution

Adherence, coherence, and aesthetic, each scored at full size where artefacts actually show. Averaged per image.

No sponsorship, nothing for sale

We don't sell the tools we rank. Outbound links go to the tool's official site and are not affiliate links.

Pricing

Leonardo uses a daily-token free tier plus paid subscription plans. Confirm current numbers on Leonardo before subscribing.

Pricing checked August 2026

Free

Daily token allowance
Access to core models
Limits and terms vary, verify on site

Apprentice

~$12/mo

Monthly token allocation
Faster generation
Commercial use

Artisan

~$30/mo

Higher token allocation
More concurrent generations
Everything in Apprentice

Maestro

~$60/mo

Top token tier
Highest limits
For heavy production use

Leonardo also bundles with Canva, and plan names and token amounts change. Verify current pricing on Leonardo directly.

Quick answers

Is Leonardo AI good for anime?

Yes, Leonardo is the best we have tested. In our benchmark its anime illustration scored 9.3, the highest of any tool, and it was the only one to render genuine soft cel shading with an authentic school uniform rather than a photo or a smooth semi-realistic look. For anime and illustration, it is a top pick.

Is Leonardo AI good for realistic images?

Very. Leonardo's photorealism scored 9.0. The fisherman portrait read the age accurately with real skin texture and honored a no-retouching brief, and its rainy street scene was the only image in our whole test to render actual falling rain. Faces and scenes are a real strength.

What is the best Leonardo model?

We tested Lucid Origin, Leonardo's current flagship, and it delivered excellent results across anime, photorealism, and product shots. Leonardo also offers an automatic mode, but it mixes different models per prompt, so for consistent output we recommend picking a single model like Lucid Origin rather than relying on automatic.

Is Leonardo AI good with hands?

This was Leonardo's weakest area, scoring 6.3. On our hand-exchange prompt it produced one person's static two-handed hold rather than a handoff, and the fingers do not hold up on close inspection. If accurate hands are essential to your image, generate a few variations or choose a tool that scored higher here.

Is Leonardo AI free?

Yes, there is a free tier with a daily token allowance and access to core models, no card required to start. Paid plans add more tokens, faster generation, and commercial use. Verify current free limits on Leonardo directly, as token amounts change.

How much does Leonardo AI cost?

The free tier is $0 with daily tokens. Paid plans run roughly $12/month (Apprentice) up to around $60/month (Maestro), with token allocations scaling by tier. Leonardo also bundles with Canva. Confirm current pricing on Leonardo, as plans change.

How did you test Leonardo AI?

We ran nine fixed prompts through Leonardo on the Lucid Origin model with automatic selection turned off, keeping the first image each time with no re-rolls. Every generator we review gets the identical nine prompts, scored on the same three-axis rubric. We sell none of the tools we rank, and we disclose exactly which model we tested.

THE BOTTOM LINE

A specialist that rewards knowing what it's for

Leonardo, on Lucid Origin, is one of the strongest creative tools we've tested, with the best anime result in our benchmark and elite photorealism and product work. But it is a specialist, not an all-rounder: it fumbles hands and follows the mood of a prompt over its precise details. Reach for it for illustration, portraits, and product shots. For legible text or exact instructions, choose more carefully.

For creators who want deep model control and beautiful output in its strong categories, Leonardo is a genuinely excellent choice, as long as you know where its edges are.

Buy it if

You want the best anime and illustration we've tested, strong photoreal and product shots, and fine control over models and styles.

Look elsewhere if

You need reliable hands, legible in-image text, or a tool that follows an exact brief over an attractive interpretation.

Other tools we tested

Same nine prompts, same rubric. The closest comparisons to this one, or browse the full list.

Pixverse output: a skincare serum bottle with a dropper on a beige background

Pixverse8.4

A video tool whose image mode excels at precision.

Read review →

Kling AI output: a split scene of desert at noon and snowy forest at night

Kling AI8.6

Strong anime, photoreal and spatial layouts.

Read review →

SeaArt output: a split scene of desert at noon and snowy forest at night

SeaArt8.0

A precise photoreal all-rounder that leads on adherence.

Read review →

See all reviews →