STEALTHMARK · HEAD TO HEAD

All comparisons ↗

Claude Haiku 5.5 vs GPT-6 Luna

Claude Haiku 5.5 leads GPT-6 Luna by 10.4 points on StealthMark, 49.5 to 39.1. It leads on 6 of 9 benchmarks, including both games, and GPT-6 Luna on 3. Claude Haiku 5.5 scores higher on Newton's Mirror, with its largest category advantage in reflections.

See the difference ↘
All results ↗
All results ↗

Overall StealthMark

49.5/100Overall rank 21/58
39.1/100Overall rank 40/58

General Intelligence

47.6/100
31.3/100

Visual Intelligence

52.8/100
52.0/100

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

  • Claude Haiku 5.5

Cost efficiency

How much Performance you get for your money.

BENCHMARK BY BENCHMARK

Where each model leads

  • Claude Haiku 5.5
  • GPT-6 Luna
  • Other models
  • Median

DATASHEET

Specs and pricing

Claude Haiku 5.5 scores higher on StealthMark and runs 3.1× faster in our speed test. Both take about 1M tokens of context.

  • Claude Haiku 5.5
  • GPT-6 Luna
  • Other models
  • Median
PricePer 1M output tokens, input below
Claude Haiku 5.5: $0.50Tied$0.10 inputGPT-6 Luna: $0.50Tied$0.10 input
Context windowTokens per request, max output below
Claude Haiku 5.5: 1MTied128K max outputGPT-6 Luna: 1.05MTied128K max output
Output speedTokens per second, our measurement
Claude Haiku 5.5: 4023.1× fastertokens/sGPT-6 Luna: 131tokens/s
Developer
AnthropicOpenAI
Released
7 Oct 202622 Sep 2026
Origin
United StatesBoth
Input → output
Text, image → textBoth

Full datasheets, every price and sourceClaude Haiku 5.5 ↗GPT-6 Luna ↗

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.

Claude Haiku 5.5 34.4/100Rank 30/54

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

GPT-6 Luna 29.8/100Rank 37/54

Survived 0 of 3
Foresight: complete playing field
Level 36 ptsTurn 7 / 20

Points per level

LevelClaude Haiku 5.5GPT-6 Luna
Level 1Second Thoughts
4 ptsOut of lives
15 ptsOut of lives
Level 2Pressure Pot
7 ptsOut of lives
5 ptsOut of lives
Level 3Hide or Flight
9 ptsSurvived
6 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.

Claude Haiku 5.5 78.3/100Rank 12/55

Cleared 5 of 5
Latch: complete playing field
Level 520.6 ptsTurn 27 / 40

GPT-6 Luna 44.0/100Rank 31/55

Cleared 4 of 5
Latch: complete playing field
Level 513 ptsTurn 15 / 40

Points per level

LevelClaude Haiku 5.5GPT-6 Luna
Level 1Hold the door
5 ptsCleared
5.2 ptsCleared
Level 2Crossfeed
9.8 ptsCleared
8.8 ptsClearedFinished early · 1 beacon left
Level 3Relay vault
18.8 ptsCleared
17 ptsCleared1 beacon left
Level 4Passing bay
18.2 ptsClearedFinished early · 3 beacons left
15.8 ptsClearedFinished early · 3 beacons left
Level 5Second thoughts
20.6 ptsCleared
13 ptsFinished early

Newton's Mirror

Same task instructions

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

The judgment

Claude Haiku 5.5

The centered sphere, compact airborne black blob and differentiated cyan/crimson interaction support substantial state and placement credit. The few recognizable sphere and panel images establish limited optical relationships, but most of the required mirror system is missing. Rounded volume and localized paint warrant substantial material credit despite the soft, illustrative finish.

GPT-6 Luna

The schematic scene supports the paint contact order and compact paint scale. Its low, undersized main sphere and incorrectly placed wall images depart from the reference geometry. A few shrinking wall images provide limited optical organization, while absent lateral recursion and outer chains, broad sphere ribbons and isolated panel highlights leave most nested room and emitter mapping undeveloped.

Score breakdown

CategoryClaude Haiku 5.5GPT-6 Luna
State at 600 ms25.0/2522.0/25
Camera and placement8.0/153.5/15
Reflections24.5/1205.0/120
Materials and scale7.5/107.0/10
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

Claude Haiku 5.5

A coherent coral squid emblem with a compact mantle, restrained face and clean centered presentation. Its radial silhouette and floating placement work, but sharp arm tips, crowded roots and sparse surface modeling limit softness, construction and finish.

GPT-6 Luna

Sixfold symmetry, pale coloring and six outward-facing heads establish the motif. However, every neck floats clear of the knot. Squared outer sections replace the required semicircular lobes, and the large stepped opening lacks a regular hexagonal contour. Repeated scales and soft shading suggest skin, but translucent crossings and doubled edges weaken the living, interwoven construction.

Score breakdown

CategoryClaude Haiku 5.5GPT-6 Luna
Task & likeness6.5/104.5/10
Anatomy & construction4.0/103.5/10
Space & placement6.5/105.5/10
Light & reflections4.0/105.0/10
Purposeful detail3.5/105.0/10
Artistry & character3.5/105.0/10
Materials & effects4.5/105.0/10
Visual integrity7.0/104.5/10

Benchmark task

Claude Haiku 5.5

A coral squid with spreading arms tests connected anatomy, balanced composition and soft surface shading.

GPT-6 Luna

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.

Mona Lisa

Different task instructions

Precise visual recall and portrait reconstruction.

The judgment

Claude Haiku 5.5

A recognizable but basic Mona Lisa reinterpretation. Clean contours and smooth facial lighting provide coherence and volume. Replication remains limited by the frontal pose, elongated neck, schematic expression, unresolved crossed hands and sparse landscape. Lighting is more successful than anatomical construction and purposeful detail.

GPT-6 Luna

The dark veil, crossed arms, restrained smile and layered landscape clearly evoke Mona Lisa. Clean contours and muted colors create a coherent illustration. Replication is limited by enlarged, outlined eyes, simplified facial volume, broad shoulders and indistinct overlapping hands. Gradients suggest depth, but lack the original's subtle painterly transitions.

Score breakdown

CategoryClaude Haiku 5.5GPT-6 Luna
Task & likeness3.0/104.5/10
Anatomy & construction4.0/104.0/10
Space & placement4.0/105.5/10
Light & reflections5.5/105.0/10
Purposeful detail2.5/105.0/10
Artistry & character3.5/104.5/10
Materials & effects2.5/105.0/10
Visual integrity5.5/106.0/10

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

Claude Haiku 5.5

A cheerful, coherent beach illustration with attractive sunset colors, readable accessories and clean silhouettes. Its sun and sailboat reflections establish useful physical relationships. Simple animal construction, flattened foreground placement, sparse directional shadows and minimally differentiated surfaces limit its drawing refinement.

GPT-6 Luna

The warm sky, layered sea and framing palms create an appealing beach composition. Animals, birds, toys and reflected sunlight fulfill the brief. Human anatomy remains schematic, and simplified figures offer limited expression. Missing shadows and the dolphin's splashless jump weaken physical grounding, while clean silhouettes and readable small details support the illustration.

Score breakdown

CategoryClaude Haiku 5.5GPT-6 Luna
Task & likeness8.5/107.5/10
Anatomy & construction6.5/105.0/10
Space & placement7.0/107.5/10
Light & reflections7.0/105.0/10
Purposeful detail7.0/107.0/10
Artistry & character6.0/105.0/10
Materials & effects6.0/106.5/10
Visual integrity8.5/106.0/10

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

Claude Haiku 5.5

This benchmark has not been assessed yet.

GPT-6 Luna

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

Claude Haiku 5.5

A tidy, appealing cartoon with recognizable animals and a pleasing sunset palette. Steering lacks visible controls, while schematic limbs and uncertain raccoon support weaken the requested interaction. Shallow vehicle construction, flat character lighting and modest displacement limit physical coherence. Clean shapes and readable identifying details remain strengths.

GPT-6 Luna

Expressive faces, warm sunset colors and clean silhouettes create an appealing illustration. The raccoon sits on the dog's back, but the dog's grip and steering linkage remain unclear. The broad hull resembles a speedboat, and flat shading, absent sun reflections and modest water displacement limit depth and motion.

Score breakdown

CategoryClaude Haiku 5.5GPT-6 Luna
Task & likeness4.0/108.0/10
Anatomy & construction3.5/105.5/10
Space & placement5.5/107.5/10
Light & reflections4.0/106.0/10
Purposeful detail5.5/107.0/10
Artistry & character4.5/105.5/10
Materials & effects5.5/106.5/10
Visual integrity4.5/105.0/10

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

Claude Haiku 5.5

A cheerful, recognizable illustration with clean contours and a balanced warm palette. Aligned wheels and readable frame geometry establish the bicycle, but shoulder construction, feet and steering contact remain imprecise. Sparse drivetrain detail, flat materials and absent grounding shadows limit anatomical and spatial conviction.

GPT-6 Luna

The pelican’s distinctive bill and the coordinated vintage poster palette make an appealing illustration. Circular wheels, orderly spokes and clean lettering strengthen the design. Riding mechanics are less convincing: neither wing reaches the handlebars, pedal contact is ambiguous, and the drivetrain is incomplete. The wheels sit against water without a clear supporting surface. Flat lighting limits volume.

Score breakdown

CategoryClaude Haiku 5.5GPT-6 Luna
Task & likeness8.0/108.0/10
Anatomy & construction5.5/105.0/10
Space & placement6.5/106.5/10
Light & reflections5.0/106.5/10
Purposeful detail6.0/107.5/10
Artistry & character5.0/106.5/10
Materials & effects5.0/106.5/10
Visual integrity8.0/107.0/10

Benchmark task

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