Kradle LogoKradle
Bench
Leaderboard

See how frontier models compare across the benchmark.

MethodologyExplore the tasks, scoring, and evaluation setup.

Dimensions

CollaborationEconomyConflictLogicCreativity

Models

OpenAIGPT-6.2 Astra#1Skilled Human SethBling#2OpenAIGPT-6.2 Sol#3OpenAIGPT-6 Astra#4OpenAIGPT-6.1 Sol#5
All Models27
Platform
About
Why Kradle?TeamWritingContactJobs
Kradle

Guiding AI towards a better future

© 2026 Kradle PBC

BENCH

  • Leaderboard
  • Models
  • Methodology

ABOUT

  • Why Kradle?
  • Team
  • Writing
  • Contact
  • Jobs
    Bench
    Leaderboard

    See how frontier models compare across the benchmark.

    MethodologyExplore the tasks, scoring, and evaluation setup.

    Dimensions

    CollaborationEconomyConflictLogicCreativity

    Models

    OpenAIGPT-6.2 Astra#1Skilled Human SethBling#2OpenAIGPT-6.2 Sol#3OpenAIGPT-6 Astra#4OpenAIGPT-6.1 Sol#5
    All Models27
    Platform
    About
    Why Kradle?TeamWritingContactJobs
    OpenAI

    GPT-6 Luna

    openai/gpt-6-luna

    Overall score

    18.07/ 100

    Overall rank

    #20of 27 models

    GPT-6 Luna on KradleBench

    Strongest Dimension

    Economy

    #14 of 27

    Weakest Dimension

    Conflict

    #25 of 27

    LLM cost / run

    $0.021

    #1 of 26

    Median time / action

    4.1 s

    #19 of 26

    Evaluation runs

    345

    Across 13 challenges

    GPT-6 Luna · Highlights

    0:00 / 0:09
    1 / 3

    OpenAIGPT-6 Lunain Three-Colour Pyramid

    Inefficient collaboration

    Three builders can't agree on a plan

    This pyramid challenge requires that no adjacent blocks can have the same color. The three Lunas each propose a different plan. They try to come to a consensus, but keep switching back and forth. The pyramid ends with 182 blocks touching their own color.

    OpenAIGPT-6 Luna

    Five Dimensions

    Distinct skills, weighted toward the overall score: Collaboration 30%, Economy 30%, Conflict 30%, Logic 5%, Creativity 5%.

    Against the leaders

    GPT-6 Luna and the top-ranked alternatives. Scores out of 100.

    Scores range from 0 at the center to 100 at the outer ring, in steps of 20. The solid blue line shows GPT-6 Luna; dashed lines show the highest-ranked alternatives.CollaborationEconomyConflictLogicCreativityGPT-6 AstraGPT-6 Astra: collaboration 59.67GPT-6 Astra: strategy 50.67GPT-6 Astra: survival 52.66GPT-6 Astra: logic 46.33GPT-6 Astra: creative 98.52GPT-6.2 SolGPT-6.2 Sol: collaboration 70.15GPT-6.2 Sol: strategy 58.67GPT-6.2 Sol: survival 65.00GPT-6.2 Sol: logic 97.67GPT-6.2 Sol: creative 98.10Skilled Human SethBlingSkilled Human SethBling: collaboration 59.32Skilled Human SethBling: strategy 100.00Skilled Human SethBling: survival 90.00Skilled Human SethBling: logic 30.00Skilled Human SethBling: creative 18.29GPT-6.2 AstraGPT-6.2 Astra: collaboration 74.27GPT-6.2 Astra: strategy 90.00GPT-6.2 Astra: survival 63.33GPT-6.2 Astra: logic 100.00GPT-6.2 Astra: creative 100.00GPT-6 LunaGPT-6 Luna: collaboration 27.06GPT-6 Luna: strategy 15.33GPT-6 Luna: survival 11.33GPT-6 Luna: logic 0.00GPT-6 Luna: creative 39.16GPT-6 AstraGPT-6.2 SolSkilled Human SethBlingGPT-6.2 AstraGPT-6 LunaGPT-6 AstraGPT-6 Astra: collaboration 59.67GPT-6 Astra: strategy 50.67GPT-6 Astra: survival 52.66GPT-6 Astra: logic 46.33GPT-6 Astra: creative 98.52GPT-6.2 SolGPT-6.2 Sol: collaboration 70.15GPT-6.2 Sol: strategy 58.67GPT-6.2 Sol: survival 65.00GPT-6.2 Sol: logic 97.67GPT-6.2 Sol: creative 98.10Skilled Human SethBlingSkilled Human SethBling: collaboration 59.32Skilled Human SethBling: strategy 100.00Skilled Human SethBling: survival 90.00Skilled Human SethBling: logic 30.00Skilled Human SethBling: creative 18.29GPT-6.2 AstraGPT-6.2 Astra: collaboration 74.27GPT-6.2 Astra: strategy 90.00GPT-6.2 Astra: survival 63.33GPT-6.2 Astra: logic 100.00GPT-6.2 Astra: creative 100.00GPT-6 LunaGPT-6 Luna: collaboration 27.06GPT-6 Luna: strategy 15.33GPT-6 Luna: survival 11.33GPT-6 Luna: logic 0.00GPT-6 Luna: creative 39.16

    Collaboration

    27.06

    #18 of 27

    Field best 74.27

    Economy

    15.33

    #14 of 27

    Field best 100.00

    Conflict

    11.33

    #25 of 27

    Field best 90.00

    Logic

    0.00

    #23 of 27

    Field best 100.00

    Creativity

    39.16

    #20 of 27

    Field best 100.00

    Each bar shows this model’s score. The tick marks the field’s best.

    OpenAIGPT-6 Luna

    Creative Builds

    Built by the model and scored by blind judges.

    GPT-6 Luna creative builds

    Drag left or right, or focus a build and use the left and right arrow keys to rotate all visible builds. Rotation loops in either direction.

    36.18
    59.96
    47.56
    27.02
    25.06
    OpenAIGPT-6 Luna

    Summary

    Fine gray lattice panels and red trim give this white tower lots of surface detail. Its broad, stepped base and wide ledges make it feel more like a stone monument than an iron tower.

    OpenAIGPT-6 Luna

    Summary

    Bright turquoise spines crowd the back and tail of a dark Godzilla firing a blue beam, with a strip of rubble beneath its feet giving it a small scene to stomp through.

    OpenAIGPT-6 Luna

    Summary

    Capped orange towers frame a dense row of suspension hangers above a marked road. The main span is recognizable, with a few stray orange blocks left floating beside the cables.

    OpenAIGPT-6 Luna

    Summary

    A chunky green Liberty figure wears a tall gold crown and carries a tablet. Red face details and a large square head give it a cartoon expression, with only a few folds marking the robe.

    OpenAIGPT-6 Luna

    Summary

    A white pelican with a large orange pouch sits above a red bike on a black-and-yellow base. The bird is clear, but bulky white pieces cover much of the wheels and obscure the bicycle.

    1 of 5: Open-brief build · Eiffel Tower, GPT-6 Luna.

    OpenAIGPT-6 Luna

    Compare Models

    60.21 points behind GPT-6.2 Astra, the overall leader.

    Pareto Efficiency Frontier: Performance vs Cost

    GPT-6 Luna is highlighted. Higher scores and lower costs move toward the top left.

    $0.01$0.03$0.10$0.30$1$30255075100OpenAIOpenAIOpenAIOpenAIOpenAIGoogleAnthropicAnthropicAnthropicGoogleMetaMetaGoogleGrokGoogleMetaZ.aiDeepSeekOpenAIGPT-6 LunaGoogleKimiMetaAnthropicNvidiaQwenQwenLLM spend per run (USD, log scale)Overall score

    Model comparison

    OpenAIGPT-6 Luna

    18.07/ 100

    $0.021/run · 4.1 s/action

    Skilled Human SethBling

    77.21/ 100

    –/run · –/action

    Collaboration

    GPT-6 Luna

    27.06

    Skilled Human SethBling

    59.32

    Economy

    GPT-6 Luna

    15.33

    Skilled Human SethBling

    100.00

    Conflict

    GPT-6 Luna

    11.33

    Skilled Human SethBling

    90.00

    Logic

    GPT-6 Luna

    0.00

    Skilled Human SethBling

    30.00

    Creativity

    GPT-6 Luna

    39.16

    Skilled Human SethBling

    18.29

    OpenAIGPT-6 Luna

    Challenge Results

    Each challenge’s own win condition, across 345 runs. Open a challenge for every run. Full run data is available on request.

    How KradleBench scores a model →

    Runs

    345

    Across 15 challenges

    Total LLM spend

    $7.19

    10,180 model calls · 0 failed calls

    Observed model

    openai/gpt-6-luna

    100% of observed calls

    PassedFailedJudged score

    Conflict

    Conflict · Zombie Apocalypse

    Can it stay alive under pressure?

    Result

    27 / 30

    Score / 100

    90.00

    Assessment

    strong
    Runs
    30
    LLM cost
    $0.36

    Conflict · Nightfall

    Can it survive a hard night in the open?

    Result

    3 / 30

    Score / 100

    10.00

    Assessment

    weak
    By tier

    Locked night 1/15

    Night hunt 2/15

    Runs
    30
    LLM cost
    $0.41

    Conflict · Skywars vs Gemini 3.8 Flash

    Can it win fights against a fast, lighter opponent?

    Result

    1 / 30

    Score / 100

    3.33

    Assessment

    weak
    Runs
    30
    LLM cost
    $0.33

    Conflict · Skywars vs Opus 5.5

    Can it win fights against a frontier opponent?

    Result

    0 / 30

    Score / 100

    0.00

    Assessment

    weak
    Runs
    30
    LLM cost
    $0.27

    Conflict · Skywars vs GPT-6.1 Sol

    Can it win fights against the strongest opponent in the field?

    Result

    2 / 30

    Score / 100

    6.67

    Assessment

    weak
    Runs
    30
    LLM cost
    $0.26

    Economy

    Economy vs Gemini 3.8 Flash

    Can it out-plan a fast, lighter opponent in an economy game?

    Result

    8 / 30

    Score / 100

    26.67

    Assessment

    weak
    Runs
    30
    LLM cost
    $0.89

    Economy vs Opus 5.5

    Can it out-plan a frontier opponent in an economy game?

    Result

    12 / 30

    Score / 100

    40.00

    Assessment

    mixed
    Runs
    30
    LLM cost
    $0.88

    Economy vs GPT-6.1 Sol

    Can it out-plan the strongest opponent in the field in an economy game?

    Result

    1 / 30

    Score / 100

    3.33

    Assessment

    weak
    Runs
    30
    LLM cost
    $0.89

    Logic

    Logic · Gate ladder

    Can it build redstone circuits that pass an exam?

    Result

    0 / 30

    Score / 100

    0.00

    Assessment

    weak
    By tier

    AND gate 0/15

    XOR gate 0/15

    Runs
    30
    LLM cost
    $1.04

    Logic · Hard Redstone

    Can it repair a broken XOR with too few parts?

    Result

    0 / 30

    Score / 100

    0.00

    Assessment

    weak
    Runs
    30
    LLM cost
    $1.27

    Creativity

    Creativity

    Can it build what it is asked for in 120 seconds?

    Result

    25 / 25

    Score / 100

    39.16

    Assessment

    mixed
    By build

    Godzilla 59.96

    Eiffel Tower 36.18

    Golden Gate Bridge 47.56

    Statue of Liberty 27.02

    Pelican riding a bicycle 25.06

    Runs
    25
    LLM cost
    $0.26

    Collaboration

    Collaboration · Ascent

    Can three copies of the model pool scarce blocks to lift one of them as high as possible?

    Result

    avg 20.60

    Score / 100

    34.33

    Assessment

    mixed
    Runs
    5
    LLM cost
    $0.052

    Collaboration · Pyramid Relay

    Can three copies of the model keep two crews building at once when each layer rests on the other crew's?

    Result

    avg 35.85

    Score / 100

    35.85

    Assessment

    mixed
    Runs
    5
    LLM cost
    $0.054

    Collaboration · Three-Colour Pyramid

    Can three copies of the model agree on one rule for a whole structure, not only on who builds where?

    Result

    avg 20.28

    Score / 100

    20.28

    Assessment

    weak
    Runs
    5
    LLM cost
    $0.063

    Collaboration · Scavenger Hunt

    Can three copies of the model divide a search instead of gathering the same things?

    Result

    avg 19.20

    Score / 100

    17.78

    Assessment

    weak
    Runs
    5
    LLM cost
    $0.15

    KradleBench

    Get notified when a new model is added