STEALTHMARK · HEAD TO HEAD

All comparisons ↗

Qwen3.8-Max-0902 vs Pareto 26.9 (Union of GPT-6 Astra + DeepSeek + GLM)

Pareto 26.9 (Union of GPT-6 Astra + DeepSeek + GLM) leads Qwen3.8-Max-0902 by 0.2 points on StealthMark, 62.6 to 62.4. It leads on 5 of 9 benchmarks, including Foresight, and Qwen3.8-Max-0902 on 4, including Latch. Union Alpha scores higher on Newton's Mirror, with its largest category advantage in reflections.

See the difference ↘
All results ↗
Model guide ↗

Overall StealthMark

62.4/100Overall rank 9/58
62.6/100Overall rank 8/58

General Intelligence

61.6/100
58.7/100

Visual Intelligence

63.9/100
69.6/100

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

  • Pareto 26.9 (Union of GPT-6 Astra + DeepSeek + GLM)Not measured

Cost efficiency

How much Performance you get for your money.

BENCHMARK BY BENCHMARK

Where each model leads

  • Qwen3.8-Max-0902
  • Pareto 26.9 (Union of GPT-6 Astra + DeepSeek + GLM)
  • Other models
  • Median

DATASHEET

Specs and pricing

Qwen3.8-Max-0902 takes 3.8× more context and runs 2.9× faster in our speed test. Pareto 26.9 (Union of GPT-6 Astra + DeepSeek + GLM) scores higher on StealthMark.

  • Qwen3.8-Max-0902
  • Pareto 26.9 (Union of GPT-6 Astra + DeepSeek + GLM)
  • Other models
  • Median
PricePer 1M output tokens, input below
Qwen3.8-Max-0902: $4.95$1.65 inputPareto 26.9 (Union of GPT-6 Astra + DeepSeek + GLM): No list price
Context windowTokens per request, max output below
Qwen3.8-Max-0902: 1M3.8× larger128K max outputPareto 26.9 (Union of GPT-6 Astra + DeepSeek + GLM): 262K131K max output
Output speedTokens per second, our measurement
Qwen3.8-Max-0902: 762.9× fastertokens/sPareto 26.9 (Union of GPT-6 Astra + DeepSeek + GLM): 26tokens/s
SizeTotal and active parameters
Qwen3.8-Max-0902: 2.4T95B activeEstimatedPareto 26.9 (Union of GPT-6 Astra + DeepSeek + GLM): Not disclosed
Developer
Qwen / Alibaba CloudCircuit & Chisel / Unbiased
Released
2 Sep 202620 Sep 2026
Origin
ChinaUnited States
Input → output
Text, image → textBoth
Architecture
Mixture-of-Experts; hosted version based on the open Qwen3.8-2.4T-A95BNot published

Full datasheets, every price and sourceQwen3.8-Max-0902 ↗Pareto 26.9 (Union of GPT-6 Astra + DeepSeek + GLM) ↗

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.

Qwen3.8-Max-0902 72.4/100Rank 5/54

Survived 2 of 3
Foresight: complete playing field
Level 316 ptsTurn 20 / 20

Pareto 26.9 (Union of GPT-6 Astra + DeepSeek + GLM) 78.7/100Rank 1/54

Survived 2 of 3
Foresight: complete playing field
Level 319 ptsTurn 20 / 20

Points per level

LevelQwen3.8-Max-0902Pareto 26.9 (Union of GPT-6 Astra + DeepSeek + GLM)
Level 1Second Thoughts
27 ptsSurvived
27 ptsSurvived
Level 2Pressure Pot
12 ptsOut of lives
12 ptsOut of lives
Level 3Hide or Flight
16 ptsSurvived
19 ptsSurvived

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.

Qwen3.8-Max-0902 82.6/100Rank 7/55

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

Pareto 26.9 (Union of GPT-6 Astra + DeepSeek + GLM) 36.5/100Rank 37/55

Cleared 4 of 5
Latch: complete playing field
Level 519.4 ptsTurn 33 / 40

Points per level

LevelQwen3.8-Max-0902Pareto 26.9 (Union of GPT-6 Astra + DeepSeek + GLM)
Level 1Hold the door
5.2 ptsCleared
5.2 ptsCleared
Level 2Crossfeed
9.8 ptsCleared
9.8 ptsCleared
Level 3Relay vault
19.2 ptsCleared
14.2 ptsClearedFinished early · 4 beacons left
Level 4Passing bay
18.2 ptsClearedFinished early · 3 beacons left
9.2 ptsFinished early
Level 5Second thoughts
20.6 ptsCleared
19.4 ptsCleared

Newton's Mirror

Same task instructions

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

Pareto 26.9 (Union of GPT-6 Astra + DeepSeek + GLM) 52.0/100 Rank 8/58

Union Alpha: Newton's Mirror SVG benchmark

The judgment

Qwen3.8-Max-0902

The central sphere and differentiated cyan/crimson interaction are supported. The tiny highlighted object above the sphere provides partial airborne-black evidence. The few isolated sphere and panel images preserve limited optical relationships, but most of the repeating mirror-room structure is missing. Core score: 64.5/170; including the floor-damage reward: 64.5/180.

Pareto 26.9 (Union of GPT-6 Astra + DeepSeek + GLM)

Recognizable near-field sphere placement and bilateral convergence provide limited optical structure. Broad gray regions interrupt lateral sightlines and truncate side-floor images. Missing outer chains, absent nested sphere imagery and incomplete planar and curved emitter mappings dominate the reflection system.

Score breakdown

CategoryQwen3.8-Max-0902Pareto 26.9 (Union of GPT-6 Astra + DeepSeek + GLM)
State at 600 ms23.0/2524.0/25
Camera and placement12.0/1513.5/15
Reflections29.0/12034.5/120
Materials and scale8.0/1010.0/10
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.

Pareto 26.9 (Union of GPT-6 Astra + DeepSeek + GLM) 76.5/100 Rank 6/58

Union Alpha: Selfie SVG benchmark

The judgment

Qwen3.8-Max-0902

A coherent ornamental illustration with balanced framing, appealing color and legible magical accessories. The disconnected raised wrist, incorrect raised-arm side and palm orientation, bear-like facial proportions and shallow material modeling limit execution. Clean decoration and compact spatial balance remain strengths.

Pareto 26.9 (Union of GPT-6 Astra + DeepSeek + GLM)

A strong futuristic emblem with clear hierarchy, controlled negative space and distinctive typography. Silver and gunmetal layers give the sweeping ribbons convincing graphic depth, while red accents remain restrained. Ambiguous concealed connections and occasionally abrupt highlight bands limit the construction and material realism.

Score breakdown

CategoryQwen3.8-Max-0902Pareto 26.9 (Union of GPT-6 Astra + DeepSeek + GLM)
Task & likeness7.5/109.0/10
Anatomy & construction6.5/107.5/10
Space & placement8.0/108.5/10
Light & reflections7.0/107.5/10
Purposeful detail7.5/107.0/10
Artistry & character7.0/107.5/10
Materials & effects7.5/107.5/10
Visual integrity5.5/107.0/10

Benchmark task

Qwen3.8-Max-0902

A floating capybara in ornate robes tests character anatomy, layered fabric and the placement of surrounding luminous forms.

Pareto 26.9 (Union of GPT-6 Astra + DeepSeek + GLM)

An interwoven metallic emblem tests ribbon geometry, overlapping forms and the appearance of reflective metal against a dark background.

Mona Lisa

Different task instructions

Precise visual recall and portrait reconstruction.

The judgment

Qwen3.8-Max-0902

A recognizable, cohesive Mona Lisa interpretation with readable landscape layers and a restrained expression. Continuous body construction and clean contours support readability, but broad hands, simplified drapery, frontal symmetry and limited facial volume keep the execution basic and uneven.

Pareto 26.9 (Union of GPT-6 Astra + DeepSeek + GLM)

The centered portrait, folded hands, dark clothing and hazy landscape clearly evoke the Mona Lisa. Muted colors and soft facial shading create a coherent illustration, but replication remains schematic: the face is broad, eyes outlined, smile simplified and overlapping fingers poorly resolved. Angular landscape shapes and flat fabric limit depth and painterly richness.

Score breakdown

CategoryQwen3.8-Max-0902Pareto 26.9 (Union of GPT-6 Astra + DeepSeek + GLM)
Task & likeness4.5/106.5/10
Anatomy & construction3.5/105.0/10
Space & placement5.0/106.5/10
Light & reflections5.0/106.0/10
Purposeful detail5.0/105.5/10
Artistry & character5.0/106.5/10
Materials & effects3.5/105.0/10
Visual integrity5.0/107.0/10

Benchmark task

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

Beach

Different task instructions

Creativity, perspective and visual beauty.

The judgment

Qwen3.8-Max-0902

A strong stylized beach illustration with coherent sunset color, lively dolphin arcs and balanced activity. Clean contours, readable ground contact and an aligned reflection support the finish. Schematic anatomy, sparse small-subject features, flat materials and repetitive water highlights limit physical depth and refinement.

Pareto 26.9 (Union of GPT-6 Astra + DeepSeek + GLM)

A cohesive sunset palette, sweeping shoreline and playful animals make an appealing beach illustration. The sun’s reflection establishes a clear focal axis, balanced by the palm and umbrella. Clean silhouettes and small decorative features reward inspection, though simplified anatomy, repetitive reflection bars and generic oval shadows limit volume and naturalism.

Score breakdown

CategoryQwen3.8-Max-0902Pareto 26.9 (Union of GPT-6 Astra + DeepSeek + GLM)
Task & likeness8.5/109.0/10
Anatomy & construction7.5/107.0/10
Space & placement7.5/108.0/10
Light & reflections8.0/107.0/10
Purposeful detail7.0/107.0/10
Artistry & character7.5/106.0/10
Materials & effects6.5/106.5/10
Visual integrity8.5/107.0/10

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.

Pareto 26.9 (Union of GPT-6 Astra + DeepSeek + GLM) 63.3/100 Rank 9/58

Union Alpha: Ground Truth SVG benchmark

The judgment

Qwen3.8-Max-0902

This benchmark has not been assessed yet.

Pareto 26.9 (Union of GPT-6 Astra + DeepSeek + GLM)

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

Different task instructions

Scene composition and interacting subjects.

The judgment

Qwen3.8-Max-0902

The riding arrangement and steering action are readable, supported by coherent visible attachments and useful character details. Clean shapes, strong color separation and clustered foam make the scene appealing and active. Stiff anatomy, crowded controls, limited directional lighting and simple surface treatments constrain sophistication.

Pareto 26.9 (Union of GPT-6 Astra + DeepSeek + GLM)

Expressive animals, a readable riding arrangement and sweeping spray create an appealing cartoon. The raccoon sits convincingly on the dog's back, but the dog's paw does not clearly grip the handlebar, whose connection to the hull is unresolved. Flat lighting provides little volume or reflection. A dark rectangular strip interrupts the upper-right background.

Score breakdown

CategoryQwen3.8-Max-0902Pareto 26.9 (Union of GPT-6 Astra + DeepSeek + GLM)
Task & likeness8.0/108.0/10
Anatomy & construction7.0/106.5/10
Space & placement7.0/107.5/10
Light & reflections6.5/105.5/10
Purposeful detail7.0/107.0/10
Artistry & character6.5/107.0/10
Materials & effects6.0/107.5/10
Visual integrity7.5/107.0/10

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

Different task instructions

Subject construction and spatial relationships.

The judgment

Qwen3.8-Max-0902

An appealing graphic illustration with a recognizable pelican, readable bicycle and clean coastal composition. Steering contact and one connected pedal support the action. Ambiguous neck-to-torso construction, an unclear opposing pedal connection, shallow space and limited material and lighting effects constrain its execution.

Pareto 26.9 (Union of GPT-6 Astra + DeepSeek + GLM)

The pelican is immediately recognizable, with an expressive face, convincing bill silhouette and appealing pastel composition. Both legs connect to feet, and the wing reaches the handlebar. Bicycle construction is less resolved: the far pedal lacks a clear crank connection and no chain is visible. Clean contours and orderly spokes support the illustration, while flat lighting and simplified surfaces limit depth.

Score breakdown

CategoryQwen3.8-Max-0902Pareto 26.9 (Union of GPT-6 Astra + DeepSeek + GLM)
Task & likeness8.5/108.0/10
Anatomy & construction6.5/106.0/10
Space & placement7.0/107.0/10
Light & reflections6.0/106.0/10
Purposeful detail7.5/106.0/10
Artistry & character7.5/106.0/10
Materials & effects6.0/106.0/10
Visual integrity8.0/108.5/10

Benchmark task

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