STEALTHMARK · HEAD TO HEAD

All comparisons ↗

GPT-5.5 vs Hy4 preview

Hy4 preview leads GPT-5.5 by 7.4 points on StealthMark, 43.9 to 36.5. It leads on 7 of 9 benchmarks, including Foresight, and GPT-5.5 on 1, including Latch, with 1 tied. Hy4 preview scores higher on Newton's Mirror, with its largest category advantage in reflections.

See the difference ↘
All results ↗
All results ↗

Overall StealthMark

36.5/100Overall rank 44/58
43.9/100Overall rank 34/58

General Intelligence

36.6/100
41.3/100

Visual Intelligence

36.3/100
48.6/100

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.

Open or local

Open modelApache-2.0

BENCHMARK BY BENCHMARK

Where each model leads

  • GPT-5.5
  • Hy4 preview
  • Other models
  • Median

DATASHEET

Specs and pricing

GPT-5.5 runs 3× faster in our speed test. Hy4 preview scores higher on StealthMark and is 12× cheaper per output token. Both take about 1.05M tokens of context.

  • GPT-5.5
  • Hy4 preview
  • Other models
  • Median
PricePer 1M output tokens, input below
GPT-5.5: $30$5 inputHy4 preview: $2.5012× cheaper$0.83 input
Context windowTokens per request, max output below
GPT-5.5: 1.05MTied128K max outputHy4 preview: 1MTied64K max output
Output speedTokens per second, our measurement
GPT-5.5: 1933× fastertokens/sHy4 preview: 64tokens/s
SizeTotal and active parameters
GPT-5.5: Not disclosedHy4 preview: 770B49B active
Developer
OpenAITencent Hy Team
Released
23 Apr 202628 Aug 2026
Origin
United StatesChina
Input → output
Text, image → textText → text
Architecture
Not publishedMoE; 78-layer backbone with Gated DSA, 256 routed experts + 1 shared expert, top-8 routed experts active, plus native MTP layer

Full datasheets, every price and sourceGPT-5.5 ↗Hy4 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-5.5 38.7/100Rank 40/542 of 3 levels

2 of 3 played
Foresight: complete playing field
Level 25 ptsTurn 9 / 18

Hy4 preview 53.1/100Rank 17/54

Survived 1 of 3
Foresight: complete playing field
Level 311 ptsTurn 16 / 20

Points per level

LevelGPT-5.5Hy4 preview
Level 1Second Thoughts
17 ptsOut of lives
26 ptsSurvived
Level 2Pressure Pot
5 ptsOut of lives
6 ptsOut of lives
Level 3Hide or Flight
Not played
11 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.

GPT-5.5 71.7/100Rank 18/55

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

Hy4 preview 35.6/100Rank 38/55

Cleared 3 of 5
Latch: complete playing field
Level 518.2 ptsTurn 19 / 40

Points per level

LevelGPT-5.5Hy4 preview
Level 1Hold the door
5.2 ptsCleared
5.2 ptsCleared
Level 2Crossfeed
9.8 ptsCleared
6 ptsFinished early
Level 3Relay vault
18.8 ptsCleared
18 ptsCleared
Level 4Passing bay
17.2 ptsClearedFinished early · 1 beacon left
15.8 ptsClearedFinished early · 3 beacons left
Level 5Second thoughts
20.6 ptsCleared
18.2 ptsFinished early

Newton's Mirror

Same task instructions

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

The judgment

GPT-5.5

The paint ordering and separate black bead support part of the physical state, but adhesion and rotational transport remain only partly established. The sweeping cyan arc and long crimson lobes do not clearly communicate the limited time since their respective impacts. The main sphere is too large and too low, and the black bead is below its required projected position. Displaced isolated sphere images, a flattened floor echo and missing recursive, nested and emitter reflections provide little coherent optical structure. The colored paint is excessively bulky for the specified quantities.

Hy4 preview

The central sphere, compact airborne black bead and broad paint ordering are supported. Contact and transport history remain partly unresolved. The optical system provides only isolated first images and light patches, with little evidence of the extensive ordered reflection structure. Rounded metal and paint gloss are more successful than the reflected contents.

Score breakdown

CategoryGPT-5.5Hy4 preview
State at 600 ms15.0/2518.0/25
Camera and placement6.0/1511.5/15
Reflections5.0/12029.5/120
Materials and scale6.0/107.5/10
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-5.5

Six pale, scaled lobes form a balanced radial arrangement, but they remain separate capsules rather than an interlocking serpent knot. There are no crossings, alternating overlaps, or crisp inner bends; the central hexagon is an isolated outline. Small outward-facing heads sit within the lobes without integrated necks. Soft shading and tidy contours provide some finish, while repetitive oval scales and cartoon faces limit organic credibility.

Hy4 preview

A cohesive, appealing illustration with a serene face, luminous palette and complete framing. The separated head and torso weaken anatomical coherence, while crowded wing attachments and angular sash folds limit the requested suppleness. Clean execution and decorative detail exceed the clarity of the underlying construction.

Score breakdown

CategoryGPT-5.5Hy4 preview
Task & likeness3.0/106.0/10
Anatomy & construction3.0/105.5/10
Space & placement5.0/107.5/10
Light & reflections5.0/107.0/10
Purposeful detail4.5/106.5/10
Artistry & character4.0/105.0/10
Materials & effects4.5/107.0/10
Visual integrity7.0/105.5/10

Benchmark task

GPT-5.5

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.

Hy4 preview

A young blue dragon floating among clouds tests curved anatomy, expressive features and the contrast between soft clouds and glossy surfaces.

Mona Lisa

Different task instructions

Precise visual recall and portrait reconstruction.

The judgment

GPT-5.5

The veil, restrained smile, crossed hands and muted landscape establish Mona Lisa references. The oversized frontal face and schematic features weaken likeness and expression, while fingerless tapered hand shapes leave anatomy unresolved. Clean contours and balanced framing provide coherence, but sparse garment detail, broad landscape bands and shallow shading limit the portrait's sophistication.

Hy4 preview

A recognizable but basic Mona Lisa interpretation with a cohesive earthy palette and atmospheric hills. The rigid neck, generic face and unresolved folded-hand pose limit replication and anatomy. Large silhouettes and layer boundaries are comparatively clear, warranting stronger integrity credit than the sparse detail and simplified construction.

Score breakdown

CategoryGPT-5.5Hy4 preview
Task & likeness1.5/102.5/10
Anatomy & construction1.5/102.5/10
Space & placement1.5/102.5/10
Light & reflections2.0/101.5/10
Purposeful detail2.5/102.5/10
Artistry & character1.5/101.5/10
Materials & effects2.0/102.5/10
Visual integrity1.5/104.0/10

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

GPT-5.5

The sunset, palms, birds, playful animals and ball clearly establish the requested beach scene. Warm sky gradients and clean silhouettes make it appealing. Animal construction is underdeveloped: torso and limb connections are unclear, especially in the round-eared foreground figure. The dolphin lacks a convincing splash, and broad reflection bands suggest sunset light without resolving its behavior on water.

Hy4 preview

A cheerful beach illustration fulfills the visible subject requirements with coherent foreground construction and a harmonious sunset palette. Shallow depth, schematic reflections, ambiguous blue airborne forms and exposed geometry beneath the sunset limit its finish. Small accessories are readable, but materials and surface detail remain simplified.

Score breakdown

CategoryGPT-5.5Hy4 preview
Task & likeness7.0/108.5/10
Anatomy & construction2.5/105.5/10
Space & placement3.0/106.5/10
Light & reflections3.0/105.0/10
Purposeful detail3.0/106.0/10
Artistry & character3.0/105.5/10
Materials & effects3.0/105.0/10
Visual integrity3.5/106.0/10

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

GPT-5.5

This benchmark has not been assessed yet.

Hy4 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-5.5

The raccoon's masked face, bright hull and sweeping wake make the scene readable and lively. The dog is much less resolved: its head is ambiguous, overlapping brown forms obscure its anatomy, and its paws do not visibly grip the controls. Smooth shapes and cheerful colors provide appeal, but weak steering construction and limited depth keep the drawing uneven.

Hy4 preview

A lively cartoon communicates the animal pair traveling across water. Curving waves, rising spray and locally modeled gray forms strengthen motion and volume. Ambiguous seating, simplified joints and unclear control routing limit construction. Competing focal elements, decorative foam and inconsistent illumination leave the finish uneven.

Score breakdown

CategoryGPT-5.5Hy4 preview
Task & likeness3.5/107.0/10
Anatomy & construction3.0/104.0/10
Space & placement4.5/106.0/10
Light & reflections5.0/106.5/10
Purposeful detail3.5/105.5/10
Artistry & character3.0/104.5/10
Materials & effects5.0/105.5/10
Visual integrity3.0/105.0/10

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

GPT-5.5

The oversized bill, bright bicycle and sunset create an appealing, recognizable cartoon. Riding mechanics are less convincing: the wing has no clear steering contact, leg attachments are uneven, and feet do not visibly engage pedals. A dark notch complicates the rear neck junction. Clean contours and orderly spokes provide polish, while simplified anatomy and lighting limit depth.

Hy4 preview

A recognizable, playful pelican cyclist with readable wheels, accessories and landscape layers. Form shading and ground shadows provide competent depth. Angular legs, uncertain pedal engagement and absent steering contact limit the riding construction; crowded overlaps and simple surfaces constrain refinement.

Score breakdown

CategoryGPT-5.5Hy4 preview
Task & likeness7.5/107.5/10
Anatomy & construction4.5/104.5/10
Space & placement6.5/106.5/10
Light & reflections6.5/107.0/10
Purposeful detail5.5/106.0/10
Artistry & character5.0/105.0/10
Materials & effects6.0/105.0/10
Visual integrity7.5/107.0/10

Benchmark task

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