STEALTHMARK · HEAD TO HEAD

All comparisons ↗

Gemma 4 31B vs Nemotron 3 Ultra

Gemma 4 31B scores 7.7 points higher overall (25.9 vs 18.2). Nemotron 3 Ultra leads General Intelligence; Gemma 4 31B leads Visual Intelligence. Gemma 4 31B scores higher on Raccoon, with its largest category advantage in visual integrity.

See the difference ↘
All results ↗
All results ↗

Overall StealthMark

25.9/100
18.2/100

General Intelligence

13.9/100
14.4/100

Visual Intelligence

40.1/100
25.7/100

Newton's Mirror

Same prompt

Physics, perspective and recursive reflections, tested together. Explore the test ↗

Gemma 4 31B15.4/100

Gemma 4 31B: Newton's Mirror SVG benchmark
Researcher observations

The sphere's volume, separate airborne black blob and vertical color ordering are readable. The large, low sphere, plain surrounding planes and absent nested optical imagery leave the central geometric and reflection requirements unsupported. Paint contact and transport remain unresolved. Material credit recognizes the visible convex shading, highlights and distinct surface colors while retaining deductions for the schematic finish and oversized blobs.

Benchmark task

A shiny steel ball spins in a room with mirrored walls and a mirrored floor. Red, cyan and black paint float above it. Gravity suddenly switches on. Draw the scene a moment later, as the ball drops and paint begins landing on its spinning surface. Show the reflections repeating through the room.

Nemotron 3 Ultra13.3/100

Nemotron 3 Ultra: Newton's Mirror SVG benchmark
Researcher observations

The candidate establishes a separate airborne black blob and partial crimson overlap, but leaves the paint interaction history and settled support unresolved. Its main silhouette, black placement and lamp projection depart substantially from the required camera geometry. A few faint reflection hints provide little support for the extensive mirror system, while the compact paint volumes are visibly disproportionate.

Benchmark task

A shiny steel ball spins in a room with mirrored walls and a mirrored floor. Red, cyan and black paint float above it. Gravity suddenly switches on. Draw the scene a moment later, as the ball drops and paint begins landing on its spinning surface. Show the reflections repeating through the room.

Compare category scores
CategoryGemma 4 31BNemotron 3 Ultra
State at 600 ms8.0/257.0/25
Camera and placement3.0/153.5/15
Reflections0.0/1206.5/120
Materials and scale6.0/103.0/10
Floor-damage reward0.0/100.0/10

Selfie

Different prompts

Gemma 4 31B60.9/100

Gemma 4 31B: Selfie SVG benchmark
Researcher observations

A clean, appealing luminous mascot with coherent body continuity and a readable jar-hugging pose. Rounded highlights, smooth gradients and balanced framing work well. The teardrop silhouette, modest eyes and nearly even grin only partly satisfy the character brief; sparse jar construction and simple geometry limit sophistication.

Prompt withheld

Prompt withheld.

Nemotron 3 Ultra36.2/100

Nemotron 3 Ultra: Selfie SVG benchmark
Researcher observations

An endearing mechanical creature is clearly framed, with expressive amber eyes and readable segmented arms. The shell resembles layered green plates more than a deep coiled spiral. Detached-looking construction, absent working contacts and patchy rendering leave the central assembly action substantially unresolved.

Prompt withheld

Prompt withheld.

Compare category scores
CategoryGemma 4 31BNemotron 3 Ultra
Prompt & likeness6.0/104.0/10
Anatomy & construction6.0/104.0/10
Space & placement7.5/105.5/10
Light & reflections6.5/105.0/10
Purposeful detail5.5/104.0/10
Artistry & character5.0/103.0/10
Materials & effects6.5/104.0/10
Visual integrity7.5/104.5/10

Mona Lisa

Different prompts

Gemma 4 31B35.1/100

Gemma 4 31B: Mona Lisa SVG benchmark
Researcher observations

The portrait is recognizable through its silhouette, palette and smile, with clean execution of simple shapes. Replication remains weak: the face and pose differ markedly, hands become fingerless ovals, clothing lacks construction, and the landscape loses its atmospheric detail. Basic gradients provide limited volume.

Benchmark task

Replicate mona-lisa as SVG

Nemotron 3 Ultra0.5/100

Nemotron 3 Ultra: Mona Lisa SVG benchmark
Researcher observations

The centered silhouette provides basic organization, but multicolored mottling and geometric overlays leave the portrait unresolved. No readable face, crossed hands, clothing or landscape establishes Mona Lisa replication. Texture and transparency offer limited visual effects without coherent anatomy, lighting or material distinctions.

Benchmark task

Replicate mona-lisa as SVG

Compare category scores
CategoryGemma 4 31BNemotron 3 Ultra
Prompt & likeness3.0/100.5/10
Anatomy & construction4.0/100.0/10
Space & placement4.0/101.0/10
Light & reflections3.5/100.5/10
Purposeful detail2.0/100.5/10
Artistry & character3.5/101.5/10
Materials & effects2.5/100.0/10
Visual integrity5.5/100.5/10

Beach

Same prompt

Gemma 4 31B29.2/100

Gemma 4 31B: Beach SVG benchmark
Researcher observations

A coherent beach illustration with clean silhouettes, a sweeping shoreline and an effective sunset palette. Requested subjects are recognizable, but primitive animal construction, sparse detail, unsupported palm placement and schematic reflections limit drawing sophistication and physical credibility.

Benchmark task

Create an SVG of a amazing beach scene. Sunset, animals, fun, birds, reflections

Nemotron 3 Ultra39.0/100

Nemotron 3 Ultra: Beach SVG benchmark
Researcher observations

A recognizable beach illustration with appealing warm-cool contrast, readable layers and mostly clean shapes. Unresolved seat construction, schematic animals and broad geometric foliage give it an elementary finish. Weakly coordinated reflections and limited material differentiation further constrain depth and atmosphere.

Benchmark task

Create an SVG of a amazing beach scene. Sunset, animals, fun, birds, reflections

Compare category scores
CategoryGemma 4 31BNemotron 3 Ultra
Prompt & likeness5.0/106.5/10
Anatomy & construction2.0/103.0/10
Space & placement5.0/105.0/10
Light & reflections2.5/101.5/10
Purposeful detail2.0/104.5/10
Artistry & character2.0/103.0/10
Materials & effects2.0/102.5/10
Visual integrity5.0/106.5/10

Ground Truth

Same prompt

Gemma 4 31B12.0/100

Gemma 4 31B: Ground Truth SVG benchmark
Benchmark task

A dark steel room has four pillars and two warm ceiling lights. A wooden plank is placed over a computer mouse, with a blueberry on the plank above it. Draw the scene a few seconds later, showing how the objects settle and how the light, shadows and surfaces look.

Nemotron 3 Ultra17.8/100

Nemotron 3 Ultra: Ground Truth SVG benchmark
Benchmark task

A dark steel room has four pillars and two warm ceiling lights. A wooden plank is placed over a computer mouse, with a blueberry on the plank above it. Draw the scene a few seconds later, showing how the objects settle and how the light, shadows and surfaces look.

Raccoon

Same prompt

Gemma 4 31B38.2/100

Gemma 4 31B: Raccoon SVG benchmark
Researcher observations

Clean, simple silhouettes communicate a raccoon riding on a dog. Limited animal specificity, absent steering controls and a flat platform weaken the requested scene. Small facial marks provide purposeful detail, while unresolved support, flat lighting and schematic water restrict physical coherence. Primitive but coherent construction supports basic drawing quality.

Benchmark task

Generate an SVG: A raccoon sits on a dog's back while the dog steers a jet ski

Nemotron 3 Ultra24.3/100

Nemotron 3 Ultra: Raccoon SVG benchmark
Researcher observations

The masked raccoon, dog and orange craft make the scene recognizable, with an appealing sunset palette. Overlapping bodies obscure the riding relationship, and the dog's connection to the controls remains unresolved. Weak water displacement, inconsistent shading and visible shape artifacts limit the finish. Ambiguous attachments do not establish detached limbs or inverted anatomy.

Benchmark task

Generate an SVG: A raccoon sits on a dog's back while the dog steers a jet ski

Compare category scores
CategoryGemma 4 31BNemotron 3 Ultra
Prompt & likeness4.5/103.5/10
Anatomy & construction4.0/101.5/10
Space & placement4.0/104.5/10
Light & reflections1.0/102.0/10
Purposeful detail3.0/103.5/10
Artistry & character4.0/101.5/10
Materials & effects2.0/103.0/10
Visual integrity8.0/102.5/10

Pelican

Same prompt

Gemma 4 31B50.5/100

Gemma 4 31B: Pelican SVG benchmark
Researcher observations

A clean, readable illustration with balanced placement and coherent basic geometry. The sky gradient and ground shadow establish diffuse illumination. Primitive shapes and sparse detail limit artistry, while unclear leg attachment, missing pedal contact and unresolved steering weaken the riding action.

Benchmark task

Generate an SVG of a pelican riding a bicycle

Nemotron 3 Ultra34.8/100

Nemotron 3 Ultra: Pelican SVG benchmark
Researcher observations

A recognizable, playful pelican dominates a readable bicycle scene. Its bill and silhouette succeed, but feet below the pedals and confused drivetrain connections weaken the riding action. Local material modeling adds interest; overlapping contours, repetitive detail and inconsistent shadows leave the illustration basic and uneven.

Benchmark task

Generate an SVG of a pelican riding a bicycle

Compare category scores
CategoryGemma 4 31BNemotron 3 Ultra
Prompt & likeness6.0/104.0/10
Anatomy & construction4.0/103.5/10
Space & placement6.0/103.5/10
Light & reflections5.0/104.0/10
Purposeful detail3.5/105.5/10
Artistry & character4.0/104.0/10
Materials & effects4.0/106.0/10
Visual integrity8.0/103.0/10