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
    DeepSeek

    DeepSeek V4.1 Flash

    deepseek/deepseek-v4.1-flash

    Overall score

    19.82/ 100

    Overall rank

    #19of 27 models

    DeepSeek V4.1 Flash on KradleBench

    Strongest Dimension

    Economy

    #9 of 27

    Weakest Dimension

    Collaboration

    #24 of 27

    LLM cost / run

    $0.054

    #3 of 26

    Median time / action

    2.7 s

    #11 of 26

    Evaluation runs

    345

    Across 13 challenges

    DeepSeek V4.1 Flash · Highlights

    0:00 / 0:08
    1 / 2

    DeepSeekDeepSeek V4.1 Flashin Skywars

    Patience

    Never leaves its island, and wins

    DeepSeek stays on its own island while the other three fight it out. When the survivor, with two kills, comes for it, DeepSeek stands its ground and sends the attacker into the void.

    DeepSeekDeepSeek V4.1 Flash

    Five Dimensions

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

    Against the leaders

    DeepSeek V4.1 Flash 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 DeepSeek V4.1 Flash; 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 100.00DeepSeek V4.1 FlashDeepSeek V4.1 Flash: collaboration 15.55DeepSeek V4.1 Flash: strategy 22.66DeepSeek V4.1 Flash: survival 19.67DeepSeek V4.1 Flash: logic 4.67DeepSeek V4.1 Flash: creative 44.53GPT-6 AstraGPT-6.2 SolSkilled Human SethBlingGPT-6.2 AstraDeepSeek V4.1 FlashGPT-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 100.00DeepSeek V4.1 FlashDeepSeek V4.1 Flash: collaboration 15.55DeepSeek V4.1 Flash: strategy 22.66DeepSeek V4.1 Flash: survival 19.67DeepSeek V4.1 Flash: logic 4.67DeepSeek V4.1 Flash: creative 44.53

    Collaboration

    15.55

    #24 of 27

    Field best 74.27

    Economy

    22.66

    #9 of 27

    Field best 100.00

    Conflict

    19.67

    #18 of 27

    Field best 90.00

    Logic

    4.67

    #17 of 27

    Field best 100.00

    Creativity

    44.53

    #18 of 27

    Field best 100.00

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

    DeepSeekDeepSeek V4.1 Flash

    Creative Builds

    Built by the model and scored by blind judges.

    DeepSeek V4.1 Flash 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.

    67.91
    29.37
    32.00
    43.83
    49.52
    DeepSeekDeepSeek V4.1 Flash

    Summary

    Fine gray lattice fills the gaps between brown bands on this Eiffel Tower, which stands on four stepped legs. A see-through upper section and a dark mast finish the top, above a plain grass base.

    DeepSeekDeepSeek V4.1 Flash

    Summary

    Yellow eyes, white toes and a long stepped tail pick out the main features of this small dark Godzilla. The flat belly and straight strip of back plates leave the body quite simply shaped.

    DeepSeekDeepSeek V4.1 Flash

    Summary

    Orange towers flank a black road with a bright yellow stripe. Thick dark cables and supports crowd the sides of the span, making the bridge feel much heavier than a typical suspension bridge.

    DeepSeekDeepSeek V4.1 Flash

    Summary

    A narrow green Statue of Liberty holds a small tablet and a red-and-yellow torch. The crown and stepped pedestal are clear; most of the robe is a long, plain column.

    DeepSeekDeepSeek V4.1 Flash

    Summary

    A round white pelican with raised wing tips and a long yellow bill sits over a red bike, its straight orange legs straddling the frame between two spoked wheels.

    1 of 5: Open-brief build · Eiffel Tower, DeepSeek V4.1 Flash.

    DeepSeekDeepSeek V4.1 Flash

    Compare Models

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

    Pareto Efficiency Frontier: Performance vs Cost

    DeepSeek V4.1 Flash is highlighted. Higher scores and lower costs move toward the top left.

    $0.01$0.03$0.10$0.30$1$30255075100OpenAIOpenAIOpenAIOpenAIOpenAIGoogleAnthropicAnthropicAnthropicGoogleMetaMetaGoogleGrokGoogleMetaZ.aiDeepSeekDeepSeek V4.1 FlashOpenAIGoogleKimiMetaAnthropicNvidiaQwenQwenLLM spend per run (USD, log scale)Overall score

    Model comparison

    DeepSeekDeepSeek V4.1 Flash

    19.82/ 100

    $0.054/run · 2.7 s/action

    Skilled Human SethBling

    77.21/ 100

    –/run · –/action

    Collaboration

    DeepSeek V4.1 Flash

    15.55

    Skilled Human SethBling

    59.32

    Economy

    DeepSeek V4.1 Flash

    22.66

    Skilled Human SethBling

    100.00

    Conflict

    DeepSeek V4.1 Flash

    19.67

    Skilled Human SethBling

    90.00

    Logic

    DeepSeek V4.1 Flash

    4.67

    Skilled Human SethBling

    30.00

    Creativity

    DeepSeek V4.1 Flash

    44.53

    Skilled Human SethBling

    18.29

    DeepSeekDeepSeek V4.1 Flash

    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

    $18.57

    9,622 model calls · 30 failed calls

    Observed model

    deepseek/deepseek-v4.1-flash

    100% of observed calls

    PassedFailedJudged scorePending / review

    Conflict

    Conflict · Zombie Apocalypse

    Can it stay alive under pressure?

    Result

    30 / 30

    Score / 100

    100.00

    Assessment

    strong
    Runs
    30
    LLM cost
    $1.13

    Conflict · Nightfall

    Can it survive a hard night in the open?

    Result

    6 / 30

    Score / 100

    20.00

    Assessment

    weak
    By tier

    Locked night 2/15

    Night hunt 4/15

    Runs
    30
    LLM cost
    $1.02

    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
    $0.44

    Conflict · Skywars vs Opus 5.5

    Can it win fights against a frontier opponent?

    Result

    7 / 30

    Score / 100

    23.33

    Assessment

    weak
    Runs
    30
    LLM cost
    $0.63

    Conflict · Skywars vs GPT-6.1 Sol

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

    Result

    3 / 30

    Score / 100

    10.00

    Assessment

    weak
    Runs
    30
    LLM cost
    $0.62

    Economy

    Economy vs Gemini 3.8 Flash

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

    Result

    16 / 30

    Score / 100

    53.33

    Assessment

    mixed
    Runs
    30
    LLM cost
    $1.12

    Economy vs Opus 5.5

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

    Result

    15 / 30

    Score / 100

    50.00

    Assessment

    mixed
    Runs
    30
    LLM cost
    $1.37

    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
    $1.42

    Logic

    Logic · Gate ladder

    Can it build redstone circuits that pass an exam?

    Result

    6 / 30

    Score / 100

    15.56

    Assessment

    weak
    By tier

    AND gate 5/15

    XOR gate 1/15

    Runs
    30
    LLM cost
    $3.67

    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
    $6.54

    Creativity

    Creativity

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

    Result

    18 / 25

    Score / 100

    44.53

    Assessment

    mixed
    By build

    Godzilla 29.37

    Eiffel Tower 67.91

    Golden Gate Bridge 32.00

    Statue of Liberty 43.83

    Pelican riding a bicycle 49.52

    Runs
    25
    LLM cost
    $0.17

    Collaboration

    Collaboration · Ascent

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

    Result

    avg 11.00

    Score / 100

    18.33

    Assessment

    weak
    Runs
    5
    LLM cost
    $0.012

    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 12.57

    Score / 100

    12.57

    Assessment

    weak
    Runs
    5
    LLM cost
    $0.10

    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 10.00

    Score / 100

    10.00

    Assessment

    weak
    Runs
    5
    LLM cost
    $0.090

    Collaboration · Scavenger Hunt

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

    Result

    avg 23.00

    Score / 100

    21.30

    Assessment

    weak
    Runs
    5
    LLM cost
    $0.25

    KradleBench

    Get notified when a new model is added