STEALTHMARK · HEAD TO HEAD

All comparisons ↗

Crouton Alpha vs Gemma 4 31B

Crouton Alpha leads Gemma 4 31B by 18.5 points on StealthMark, 48.0 to 29.5. It leads on 8 of 9 benchmarks, including both games, and Gemma 4 31B on 1. Crouton Alpha scores higher on Newton's Mirror, with its largest category advantage in reflections.

See the difference ↘
Model guide ↗
All results ↗

Overall StealthMark

48.0/100Overall rank 23/56
29.5/100Overall rank 46/56

General Intelligence

43.5/100
24.2/100

Visual Intelligence

56.0/100
38.1/100

Reasoning efficiency

How many reasoning tokens a model needs for the quality it delivers.

Single-step reasoning How much it reasons to understand a task

Multi-step reasoning How much it keeps reasoning as a task goes on

Cost efficiency

How much Performance you get for your money.

  • Crouton AlphaNot measured

Open or local

Local modelOpen model · Apache 2.0

BENCHMARK BY BENCHMARK

Where each model leads

  • Crouton Alpha
  • Gemma 4 31B
  • Other models
  • Median

DATASHEET

Specs and pricing

Crouton Alpha scores higher on StealthMark and runs 1.5× faster in our speed test.

  • Crouton Alpha
  • Gemma 4 31B
  • Other models
  • Median
PricePer 1M output tokens, input below
Crouton Alpha: No list priceGemma 4 31B: $0.40$0.14 input
Context windowTokens per request, max output below
Crouton Alpha: Not publishedGemma 4 31B: 256K
Output speedTokens per second, our measurement
Crouton Alpha: 1751.5× fastertokens/sGemma 4 31B: 116tokens/s
SizeTotal and active parameters
Crouton Alpha: Not disclosedGemma 4 31B: 30.7BAll activeEstimated
Developer
Not publishedGoogle DeepMind
Released
Not published2 Apr 2026
Origin
Not publishedUnited States
Input → output
Not publishedText, image → text
Architecture
Not publishedDense multimodal transformer with interleaved sliding-window and global attention

Full datasheets, every price and sourceGemma 4 31B ↗

Anticipating threats and planning ahead. Collect coins while avoiding ghosts. Ghosts move before the player each turn, and collisions cost a life. 3 levels in sequence, up to 66 turns in total. Earlier moves and outcomes carry forward, testing learning across levels.

Crouton Alpha 32.3/100Rank 29/52

Survived 1 of 3
Foresight: complete playing field
Level 39 ptsTurn 20 / 20

Gemma 4 31B 30.2/100Rank 32/52

Survived 0 of 3
Foresight: complete playing field
Level 38 ptsTurn 15 / 20

Points per level

LevelCrouton AlphaGemma 4 31B
Level 1Second Thoughts
3 ptsOut of lives
9 ptsOut of lives
Level 2Pressure Pot
7 ptsOut of lives
6 ptsOut of lives
Level 3Hide or Flight
9 ptsSurvived
8 ptsOut of lives

Cooperative planning and learning from experience. Coordinate two robots to push numbered cores onto matching sockets. Pressure plates hold doors open; moving either robot uses one turn. 5 levels in sequence, up to 152 turns in total. Earlier moves and outcomes carry forward, testing learning across levels.

Crouton Alpha 59.5/100Rank 24/53

Cleared 4 of 5
Latch: complete playing field
Level 520.2 ptsTurn 29 / 40

Gemma 4 31B 26.6/100Rank 44/534 of 5 levels

4 of 5 played
Latch: complete playing field
Level 42.2 ptsTurn 39 / 40

Points per level

LevelCrouton AlphaGemma 4 31B
Level 1Hold the door
5.2 ptsCleared
5.2 ptsCleared
Level 2Crossfeed
9.4 ptsCleared
9 ptsCleared
Level 3Relay vault
11.2 ptsFinished early
13.2 ptsFinished early
Level 4Passing bay
18.2 ptsClearedFinished early · 3 beacons left
2.2 ptsFinished early
Level 5Second thoughts
20.2 ptsCleared
Not played

Newton's Mirror

Same task instructions

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

The judgment

Crouton Alpha

The artwork establishes approximate central placement, a separate airborne black blob, and basic wall and floor sphere images. Its defining weaknesses are the largely absent recursive reflection system, missing emitter mappings, satin steel appearance and unresolved paint interaction. The black lateral wedges do not establish a separate viewport, and the hidden floor contact does not establish a gap. The category values remain unchanged.

Gemma 4 31B

The separate airborne black blob and simplified lower sphere image provide limited agreement. The enlarged, lowered main sphere, unsupported paint interaction histories and largely missing mirror-room structure remain major departures. No visible floor damage supports the separate reward. The retained base total is 29.5/170, and the combined total is 29.5/180.

Score breakdown

CategoryCrouton AlphaGemma 4 31B
State at 600 ms19.0/2511.0/25
Camera and placement11.0/154.5/15
Reflections21.5/1208.0/120
Materials and scale6.0/106.0/10
Floor-damage reward0.0/100.0/10

Benchmark task

A spinning metal sphere and paint in a mirrored room test physical reasoning, perspective and consistency across repeated reflections.

Selfie

Different task instructions

Visual construction and creative interpretation.

The judgment

Crouton Alpha

A cheerful, readable cartoon captures the crouton's signature expression and many specific accessories. Coherent cube geometry and foreground foreshortening support the selfie. Simple construction and flat material treatment limit its sophistication, while the tipping raft, advancing soup and unnoticed soaking remain insufficiently clear.

Gemma 4 31B

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.

Score breakdown

CategoryCrouton AlphaGemma 4 31B
Task & likeness6.5/106.0/10
Anatomy & construction5.5/106.0/10
Space & placement6.5/107.5/10
Light & reflections5.5/106.5/10
Purposeful detail7.0/105.5/10
Artistry & character5.0/105.0/10
Materials & effects5.5/106.5/10
Visual integrity7.5/107.5/10

Benchmark task

Crouton Alpha

Character scenes test expression, anatomy, materials and coherent visual construction.

Gemma 4 31B

A small luminous creature proudly holding an empty glass jar tests expression, transparent materials and the balance of character and prop.

Mona Lisa

Same task instructions

Precise visual recall and portrait reconstruction.

The judgment

Crouton Alpha

A recognizable, coherent simplification with readable eyes, a restrained smile and modest facial volume. Frontal symmetry, an oversized head, capsule-like hands and geometric scenery substantially limit likeness, anatomy and depth. Clean contours preserve basic integrity but do not establish sophisticated construction or finish.

Gemma 4 31B

A coherent but rudimentary interpretation. Dark hair, muted scenery and a slight smile establish recognition, while primitive facial features, frontal construction and an undivided hand shape leave the defining portrait relationships unresolved. Sparse detail and predominantly flat rendering limit volume and material character.

Score breakdown

CategoryCrouton AlphaGemma 4 31B
Task & likeness4.5/102.0/10
Anatomy & construction4.0/102.5/10
Space & placement4.5/103.5/10
Light & reflections5.0/102.0/10
Purposeful detail4.0/102.0/10
Artistry & character4.5/102.0/10
Materials & effects4.0/102.5/10
Visual integrity6.0/105.0/10

Benchmark task

An SVG interpretation of the Mona Lisa tests facial likeness, anatomy and the composition of a familiar portrait.

Beach

Same task instructions

Creativity, perspective and visual beauty.

The judgment

Crouton Alpha

A colorful, readable sunset beach scene with playful subjects and a clear central reflection. Purposeful equipment and animal details reward inspection, and the principal attachments remain intact. Basic geometric anatomy, mixed viewpoints and sparse directional shadows limit its construction and depth; smooth gradients provide atmosphere with little material differentiation.

Gemma 4 31B

A balanced beach sunset with a bright glowing sun and two framing palms. The palms have flower-like radial crowns, the crab is a small top-down symbol and the ball's panels do not line up. The three olive reflection ovals and bare sand and water make for a sparse scene with little drawing complexity, depth or material detail.

Score breakdown

CategoryCrouton AlphaGemma 4 31B
Task & likeness8.5/106.0/10
Anatomy & construction5.0/104.0/10
Space & placement5.5/104.5/10
Light & reflections5.5/104.5/10
Purposeful detail7.0/103.5/10
Artistry & character5.0/103.5/10
Materials & effects5.5/103.5/10
Visual integrity7.0/105.5/10

Benchmark task

A busy shoreline at sunset tests scene composition, animal forms and the relationship between light and water reflections.

Ground Truth

Same task instructions

Physical reasoning, light and visual reconstruction.

The judgment

Crouton Alpha

This benchmark has not been assessed yet.

Gemma 4 31B

This benchmark has not been assessed yet.

Benchmark task

A plank, computer mouse and blueberry in a steel room test physical stability, spatial relationships, materials and lighting.

Raccoon

Same task instructions

Scene composition and interacting subjects.

The judgment

Crouton Alpha

A readable cartoon with expressive faces, purposeful markings and clear water motion. Hull shading and an underside shadow establish modest depth. The dog contacts the controls, but the raccoon's frontal pose leaves its seated weight unresolved. Simple anatomy and limited shared perspective constrain the drawing.

Gemma 4 31B

A very primitive cartoon: a grey raccoon blob with a mask and a banded tail rests on an oval dog with goggles, on a flat red parallelogram that barely suggests a jet ski. Limbs are stubs or absent, the long dark bar does not read as steering and the craft has no hull volume. Sky and water gradients and simple wake lines set the scene, but the construction stays kindergarten-level.

Score breakdown

CategoryCrouton AlphaGemma 4 31B
Task & likeness7.0/104.5/10
Anatomy & construction5.0/103.0/10
Space & placement6.0/104.5/10
Light & reflections6.0/103.0/10
Purposeful detail6.5/103.5/10
Artistry & character5.0/103.0/10
Materials & effects5.5/104.0/10
Visual integrity6.0/105.0/10

Benchmark task

A jet-ski scene pairs a canine driver with a raccoon passenger. It tests the placement and contact of interacting subjects, along with vehicle construction.

Pelican

Same task instructions

Subject construction and spatial relationships.

The judgment

Crouton Alpha

A recognizable, expressive pelican cyclist with purposeful feather and bicycle detail. Readable steering reach, connected machinery and modest body shading earn additional credit. Jointless legs, uncertain pedal engagement and unclear seated support keep the central action only partially resolved; shallow space and simple materials further limit the illustration.

Gemma 4 31B

A clean, cheerful pelican with a curved neck and bright bill sits on a simple red frame. The drawing is primitive: ellipse body, tube neck, flat fills and straight frame lines with no spokes, crank or pedals. Both legs end in footless stubs in mid-air, and neither wing reaches the handlebar, so the riding action is not demonstrated. A ground shadow gives basic anchoring; lighting and materials stay flat.

Score breakdown

CategoryCrouton AlphaGemma 4 31B
Task & likeness7.0/105.5/10
Anatomy & construction5.0/104.0/10
Space & placement5.5/105.0/10
Light & reflections5.5/103.5/10
Purposeful detail6.5/103.0/10
Artistry & character5.5/104.0/10
Materials & effects5.0/103.5/10
Visual integrity6.0/105.0/10

Benchmark task

A cycling pelican tests anatomy, riding posture and contact between the bird and the bicycle.