STEALTHMARK · HEAD TO HEAD

All comparisons ↗

Gemma 4 31B vs Step 5 Preview

Step 5 Preview leads Gemma 4 31B by 13.2 points on StealthMark, 39.0 to 25.8. It leads on 6 of 9 benchmarks, including Latch, and Gemma 4 31B on 3, including Foresight. Step 5 Preview scores higher on Newton's Mirror, with its largest category advantage in reflections.

See the difference ↘
All results ↗
All results ↗

Overall StealthMark

25.8/100Overall rank 44/53
39.0/100Overall rank 36/53

General Intelligence

19.1/100
37.1/100

Visual Intelligence

36.7/100
41.9/100

Reasoning efficiency

On every task, result quality and reasoning tokens are each compared with all measured models and weighed equally. Each dot is one model.

Single-step reasoning Result quality weighed against reasoning tokens on a single problem

Multi-step reasoning Result quality weighed against reasoning tokens across connected steps, as in agentic workflows

Open or local

Local modelOpen model · Apache 2.0

BENCHMARK BY BENCHMARK

Where each model leads

  • Gemma 4 31B
  • Step 5 Preview
  • Other models
  • Median

DATASHEET

Specs and pricing

Step 5 Preview scores higher on StealthMark and takes 4× more context.

  • Gemma 4 31B
  • Step 5 Preview
  • Other models
  • Median
PricePer 1M output tokens, input below
Gemma 4 31B: $0.40$0.14 inputStep 5 Preview: No list price
Context windowTokens per request, max output below
Gemma 4 31B: 256KStep 5 Preview: 1M4× larger
Output speedTokens per second, our measurement
Gemma 4 31B: 116tokens/sStep 5 Preview: Not measured
SizeTotal and active parameters
Gemma 4 31B: 30.7BAll activeEstimatedStep 5 Preview: 600B27B active
Developer
Google DeepMindStepFun
Released
2 Apr 2026Not published
Origin
United StatesNot published
Input → output
Text, image → textText, image, video → text
Architecture
Dense multimodal transformer with interleaved sliding-window and global attentionSparse mixture-of-experts; 92 layers (88 MoE), 352 routed experts with 8 selected per token plus a shared expert; grouped-query attention with 512-token sliding windows and full attention every fourth layer; 3 multi-token-prediction layers; vision tower for image and video input

Full datasheets, every price and sourceGemma 4 31B ↗Step 5 Preview ↗

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.

Gemma 4 31B 30.2/100Rank 29/49

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

Step 5 Preview 59.9/100Rank 42/491 of 3 levels

1 of 3 played
Foresight: complete playing field
Level 122 ptsTurn 28 / 28

Points per level

LevelGemma 4 31BStep 5 Preview
Level 1Second Thoughts
9 ptsOut of lives
22 ptsSurvived
Level 2Pressure Pot
6 ptsOut of lives
Pending
Level 3Hide or Flight
8 ptsOut of lives
Pending

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.

Gemma 4 31B 26.6/100Rank 41/504 of 5 levels

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

Step 5 Preview 40.7/100Rank 28/50

Cleared 4 of 5
Latch: complete playing field
Level 515 ptsTurn 35 / 40

Points per level

LevelGemma 4 31BStep 5 Preview
Level 1Hold the door
5.2 ptsCleared
5.2 ptsCleared
Level 2Crossfeed
9 ptsCleared
9 ptsCleared
Level 3Relay vault
13.2 ptsFinished early
18.6 ptsCleared
Level 4Passing bay
2.2 ptsFinished early
15.6 ptsClearedFinished early · 3 beacons left
Level 5Second thoughts
Not played
15 ptsFinished early

Newton's Mirror

Same task instructions

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

The judgment

Gemma 4 31B

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.

Step 5 Preview

The main sphere, airborne black and small colored features occupy broadly appropriate positions. Most of the defining mirror system is absent, and paint interaction remains only partially supported. The small paint features do not visibly establish a separate volume-scale violation.

Score breakdown

CategoryGemma 4 31BStep 5 Preview
State at 600 ms11.0/2520.0/25
Camera and placement3.0/1513.5/15
Reflections0.0/12020.0/120
Materials and scale5.5/107.0/10
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

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.

Step 5 Preview

A charming diagonal climber and warm palette communicate the central idea. Simple rounded construction, crowded limb counting, uncertain footholds and an unattached upper shelf limit its physical coherence. Growth bands and the yarn trail add specificity, but the requested material distinctions and unravelled bootie remain underdeveloped.

Score breakdown

CategoryGemma 4 31BStep 5 Preview
Task & likeness6.0/106.0/10
Anatomy & construction6.0/104.5/10
Space & placement7.5/105.0/10
Light & reflections6.5/105.5/10
Purposeful detail5.5/105.5/10
Artistry & character5.0/105.0/10
Materials & effects6.5/106.0/10
Visual integrity7.5/105.0/10

Benchmark task

Gemma 4 31B

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

Step 5 Preview

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

Mona Lisa

Same task instructions

Precise visual recall and portrait reconstruction.

The judgment

Gemma 4 31B

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.

Step 5 Preview

A recognizable, coherent simplification with purposeful facial contours and differentiated fingers. These provide more construction and detail than the broad silhouette alone suggests. The oversized face, frontal seated pose, unresolved wrists and sparse landscape still substantially limit likeness, depth and finish.

Score breakdown

CategoryGemma 4 31BStep 5 Preview
Task & likeness3.0/104.0/10
Anatomy & construction4.0/104.0/10
Space & placement4.0/104.0/10
Light & reflections3.5/103.5/10
Purposeful detail2.0/104.0/10
Artistry & character3.5/104.0/10
Materials & effects2.5/103.0/10
Visual integrity5.5/103.5/10

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

Gemma 4 31B

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.

Step 5 Preview

A recognizable sunset beach with clear warm-cool contrast and playful accessories. Basic animal construction, shallow foreground placement and schematic reflections limit spatial and physical credibility. Sparse descriptive detail and abruptly ended sky bands constrain the finish.

Score breakdown

CategoryGemma 4 31BStep 5 Preview
Task & likeness5.0/104.0/10
Anatomy & construction2.0/103.0/10
Space & placement5.0/103.5/10
Light & reflections2.5/103.5/10
Purposeful detail2.0/103.0/10
Artistry & character2.0/103.0/10
Materials & effects2.0/103.5/10
Visual integrity5.0/104.0/10

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

Gemma 4 31B

This benchmark has not been assessed yet.

Step 5 Preview

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

Gemma 4 31B

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.

Step 5 Preview

Recognizable animals, a bright palette and identifying accessories support a basic cartoon. Primitive anatomy, partial seated support, unresolved steering and weak hull-water contact limit the central action. Lighting and spray provide atmosphere without establishing convincing physical depth.

Score breakdown

CategoryGemma 4 31BStep 5 Preview
Task & likeness2.5/104.0/10
Anatomy & construction3.0/104.5/10
Space & placement4.0/103.0/10
Light & reflections1.0/103.5/10
Purposeful detail2.5/104.0/10
Artistry & character2.5/103.5/10
Materials & effects2.0/103.0/10
Visual integrity7.0/103.0/10

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

Gemma 4 31B

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.

Step 5 Preview

The pelican and bicycle are recognizable, with a pleasant palette and basic diffuse shading. The riding relationship fails physically: the bird is unsupported, the front wheel is unmounted, and pedalling is unresolved. Simple geometric construction and sparse material treatment further limit the drawing.

Score breakdown

CategoryGemma 4 31BStep 5 Preview
Task & likeness4.0/103.0/10
Anatomy & construction3.0/103.0/10
Space & placement4.5/103.0/10
Light & reflections3.0/103.0/10
Purposeful detail3.0/104.0/10
Artistry & character3.0/103.0/10
Materials & effects4.0/103.0/10
Visual integrity4.5/102.0/10

Benchmark task

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