STEALTHMARK · HEAD TO HEAD

All comparisons ↗

GPT-6 Astra vs Step 5 Preview

GPT-6 Astra leads Step 5 Preview by 35.7 points on StealthMark, 74.7 to 39.0. It leads on all 9 benchmarks. GPT-6 Astra scores higher on Newton's Mirror, with its largest category advantage in reflections.

See the difference ↘
All results ↗
All results ↗

Overall StealthMark

74.7/100Overall rank 1/53
39.0/100Overall rank 36/53

General Intelligence

73.4/100
37.1/100

Visual Intelligence

77.0/100
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

  • GPT-6 Astra

BENCHMARK BY BENCHMARK

Where each model leads

  • GPT-6 Astra
  • Step 5 Preview
  • Other models
  • Median

DATASHEET

Specs and pricing

Both take about 1.05M tokens of context.

  • GPT-6 Astra
  • Step 5 Preview
  • Other models
  • Median
PricePer 1M output tokens, input below
GPT-6 Astra: $50$10 inputStep 5 Preview: No list price
Context windowTokens per request, max output below
GPT-6 Astra: 1.05MTied128K max outputStep 5 Preview: 1MTied
Output speedTokens per second, our measurement
GPT-6 Astra: 65tokens/sStep 5 Preview: Not measured
SizeTotal and active parameters
GPT-6 Astra: Not disclosedStep 5 Preview: 600B27B active
Developer
OpenAIStepFun
Released
3 Sep 2026Not published
Origin
United StatesNot published
Input → output
Text, image → textText, image, video → text
Architecture
Not publishedSparse 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 sourceGPT-6 Astra ↗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.

GPT-6 Astra 76.1/100Rank 2/49

Survived 3 of 3
Foresight: complete playing field
Level 318 ptsTurn 20 / 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

LevelGPT-6 AstraStep 5 Preview
Level 1Second Thoughts
27 ptsSurvived
22 ptsSurvived
Level 2Pressure Pot
11 ptsSurvived
Pending
Level 3Hide or Flight
18 ptsSurvived
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.

GPT-6 Astra 87.6/100Rank 1/50

Cleared 5 of 5
Latch: complete playing field
Level 520.6 ptsTurn 27 / 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

LevelGPT-6 AstraStep 5 Preview
Level 1Hold the door
5.2 ptsCleared
5.2 ptsCleared
Level 2Crossfeed
10.2 ptsClearedFinished early · 1 beacon left
9 ptsCleared
Level 3Relay vault
19.2 ptsCleared
18.6 ptsCleared
Level 4Passing bay
18.2 ptsClearedFinished early · 3 beacons left
15.6 ptsClearedFinished early · 3 beacons left
Level 5Second thoughts
20.6 ptsCleared
15 ptsFinished early

Newton's Mirror

Same task instructions

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

The judgment

GPT-6 Astra

Major placement and the visible contact sequence are convincing. Dense planar light repetition, recognizable spherical panel images and simplified nested sphere motifs provide useful optical structure. Missing outer chains, truncated distant sequences and incorrect bead-like belts still leave the recursive reflections substantially incomplete.

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

CategoryGPT-6 AstraStep 5 Preview
State at 600 ms25.0/2520.0/25
Camera and placement15.0/1513.5/15
Reflections70.0/12020.0/120
Materials and scale10.0/107.0/10
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

GPT-6 Astra

The sixfold silhouette, pale coloring, outward heads and central hexagon closely match the brief. Fine scales and soft shading support rounded volume. However, several crossings have conspicuous rectangular seams and weak contact shadows, making the weave feel assembled. Small heads have limited anatomical definition, and the repetitive texture conveys less organic life than requested.

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

CategoryGPT-6 AstraStep 5 Preview
Task & likeness9.0/106.0/10
Anatomy & construction8.0/104.5/10
Space & placement7.5/105.0/10
Light & reflections7.0/105.5/10
Purposeful detail8.0/105.5/10
Artistry & character8.5/105.0/10
Materials & effects8.0/106.0/10
Visual integrity8.5/105.0/10

Benchmark task

GPT-6 Astra

A pale serpent woven into an interlocking knot tests connected anatomy, repeated structure and readable overlaps. Organic surface detail must work with the overall geometry.

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

GPT-6 Astra

The crossed hands, restrained smile, dark clothing and hazy landscape clearly recall the Mona Lisa. Replication is limited by the frontal, elongated face, schematic eyes and stiff fingers. Soft shading gives basic volume, but broad garment shapes and angular scenery lack the original’s nuanced depth and painterly transitions. The frame and composition are controlled; surface grain adds texture without resolving the simplified drawing.

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

CategoryGPT-6 AstraStep 5 Preview
Task & likeness6.5/104.0/10
Anatomy & construction6.5/104.0/10
Space & placement6.5/104.0/10
Light & reflections6.5/103.5/10
Purposeful detail6.5/104.0/10
Artistry & character7.0/104.0/10
Materials & effects7.0/103.0/10
Visual integrity7.5/103.5/10

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

GPT-6 Astra

A richly composed sunset beach illustration with cohesive peach, mauve and teal colors. Sweeping palm fronds, a scalloped umbrella and curved shorelines frame the sun and its aligned reflection. Distinct animal silhouettes and carefully distributed beach details reward inspection. Simplified human and dog anatomy, slight dolphin splash cues and inconsistent equipment shadows limit physical conviction.

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

CategoryGPT-6 AstraStep 5 Preview
Task & likeness9.0/104.0/10
Anatomy & construction7.5/103.0/10
Space & placement8.5/103.5/10
Light & reflections7.5/103.5/10
Purposeful detail8.5/103.0/10
Artistry & character8.5/103.0/10
Materials & effects7.5/103.5/10
Visual integrity7.5/104.0/10

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

GPT-6 Astra

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

GPT-6 Astra

Expressive animals, a balanced sunset composition and sweeping wake make an appealing illustration. The raccoon rides the dog's back, and the dog's connected foreleg reaches the controls. The compact steering assembly and layered paws are less clearly constructed. Lighting remains mostly flat, with little shadow or sun reflection. Clean contours and purposeful details support strong illustrative execution.

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

CategoryGPT-6 AstraStep 5 Preview
Task & likeness8.5/104.0/10
Anatomy & construction7.5/104.5/10
Space & placement8.0/103.0/10
Light & reflections6.5/103.5/10
Purposeful detail8.0/104.0/10
Artistry & character7.5/103.5/10
Materials & effects8.0/103.0/10
Visual integrity8.5/103.0/10

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

GPT-6 Astra

The pelican's long bill, flowing scarf and warm coastal palette create a charming, balanced illustration. Wheels, spokes and frame are largely coherent, but neither wing reaches the handlebars, and the rear leg's pedal contact is unclear. Flat lighting and simplified surfaces limit depth despite clean outlines and attractive feather detail.

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

CategoryGPT-6 AstraStep 5 Preview
Task & likeness8.0/103.0/10
Anatomy & construction7.0/103.0/10
Space & placement7.5/103.0/10
Light & reflections7.0/103.0/10
Purposeful detail8.0/104.0/10
Artistry & character8.0/103.0/10
Materials & effects7.5/103.0/10
Visual integrity8.0/102.0/10

Benchmark task

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