OpenAI: GPT-4o (2024-08-06)

The 2024-08-06 version of GPT-4o offers improved performance in structured outputs, with the ability to supply a JSON schema in the respone_format. Read more [here](https://openai.com/index/introduci…

openai/gpt-4o-2024-08-06

The 2024-08-06 version of GPT-4o offers improved performance in structured outputs, with the ability to supply a JSON schema in the respone_format. Read more [here](https://openai.com/index/introducing-structured-outputs-in-the-api/). GPT-4o ("o" for "omni") is...

Specifications & capabilities

Input
text, image, file
Output
text
Context window · tokens
128000
Knowledge cutoff
2023-10-31
Capabilities
fileInput, multimodal, promptCaching, structuredOutputs, toolCalling, visionInput

Evaluations

61–80 / 143

Kaggle Benchmarks

36.8%

source id
kaggleBenchmarks
leaderboard name
MultiLoKo Hindi · EM Hindi
dimension
reasoning
metric
MultiLoKo Hindi · EM Hindi
rank
13
aggregation scope
kaggle
true
status
published
license
Apache-2.0
direction
higherIsBetter
conditions
task
id
39
name
EM Hindi
version
1
sort_order
DESCENDING
display_type
percentages
benchmark
id
62
name
MultiLoKo Hindi
type
INDIVIDUAL
owner
metaresearch
version
1
categories
  • multilingual and cross-lingual capabilities
  • general knowledge and reasoning
version_id
48
maintainer_tier
Kaggle-maintained
maintenance_level
KAGGLE_MAINTAINED
candidate
candidate_class
modelVersion
record_kind
aggregate
attribution
Kaggle · metaresearch · MultiLoKo Hindi
license_url
https://www.apache.org/licenses/LICENSE-2.0
metric_kind
MultiLoKo Hindi · EM Hindi
rank_method
competition rank among reported evaluations in identical conditions, including unlinked observations
source_file
48/leaderboard.json
source_page
https://www.kaggle.com/benchmarks/metaresearch/multiloko-hindi/versions/1
measurements
  • name
    EM Hindi
    unit
    percent
    scale
    1
    metric
    MultiLoKo Hindi · EM Hindi
    result
    resultCase
    numericResult
    booleanResult
    false
    numericResult
    value
    0.368
    standardDeviation
    0
    confidenceInterval
    0.0597806365533776
    hasStandardDeviation
    false
    hasConfidenceInterval
    true
    confidenceIntervalNullable
    0.0597806365533776
    hasUnevenConfidenceInterval
    false
    taskVersionId
    0
    evaluationDate
    2025-06-16T00:00:00Z
    hasBooleanResult
    false
    hasNumericResult
    true
    hasTaskVersionId
    false
    hasEvaluationDate
    true
    numericResultNullable
    value
    0.368
    standardDeviation
    0
    confidenceInterval
    0.0597806365533776
    hasStandardDeviation
    false
    hasConfidenceInterval
    true
    confidenceIntervalNullable
    0.0597806365533776
    hasUnevenConfidenceInterval
    false
    hasNumericResultPublic
    false
    customAdditionalResults
      hasNumericResultPrivate
      false
      task_id
      39
      sort_order
      DESCENDING
      source_name
      EM Hindi
      display_type
      percentages
      task_version
      1
      task_display_type
      PERCENTAGES
      normalization_status
      normalized
    published_at
    2025-07-15T11:49:31.790Z
    candidate_class
    modelVersion
    source_revision
    b32d007df5acbddafca23c7dd2beba8221c519cf7dd3445a1d7bc28740ffd0d8
    publisher_anchor
    openai/gpt-4o-2024-08-06
    metric_population
    51
    benchmark_evaluation
    true
    task_metadata_source
    benchmarkVersion
    candidate_metadata_source
    benchmarkVersionMappings
    evaluated at
    2025-06-16T00:00:00.000Z
    source model id
    GPT-4o
    confidence lower
    0.3082193634466224
    confidence upper
    0.4277806365533776
    Original result ↗

    Kaggle Benchmarks

    29.2%

    source id
    kaggleBenchmarks
    leaderboard name
    MultiLoKo Marathi · EM Marathi
    dimension
    reasoning
    metric
    MultiLoKo Marathi · EM Marathi
    rank
    13
    aggregation scope
    kaggle
    true
    status
    published
    license
    Apache-2.0
    direction
    higherIsBetter
    conditions
    task
    id
    46
    name
    EM Marathi
    version
    1
    sort_order
    DESCENDING
    display_type
    percentages
    benchmark
    id
    69
    name
    MultiLoKo Marathi
    type
    INDIVIDUAL
    owner
    metaresearch
    version
    1
    categories
    • multilingual and cross-lingual capabilities
    • general knowledge and reasoning
    version_id
    55
    maintainer_tier
    Kaggle-maintained
    maintenance_level
    KAGGLE_MAINTAINED
    candidate
    candidate_class
    modelVersion
    record_kind
    aggregate
    attribution
    Kaggle · metaresearch · MultiLoKo Marathi
    license_url
    https://www.apache.org/licenses/LICENSE-2.0
    metric_kind
    MultiLoKo Marathi · EM Marathi
    rank_method
    competition rank among reported evaluations in identical conditions, including unlinked observations
    source_file
    55/leaderboard.json
    source_page
    https://www.kaggle.com/benchmarks/metaresearch/multiloko-marathi/versions/1
    measurements
    • name
      EM Marathi
      unit
      percent
      scale
      1
      metric
      MultiLoKo Marathi · EM Marathi
      result
      resultCase
      numericResult
      booleanResult
      false
      numericResult
      value
      0.292
      standardDeviation
      0
      confidenceInterval
      0.0563619669903391
      hasStandardDeviation
      false
      hasConfidenceInterval
      true
      confidenceIntervalNullable
      0.0563619669903391
      hasUnevenConfidenceInterval
      false
      taskVersionId
      0
      evaluationDate
      2025-06-16T00:00:00Z
      hasBooleanResult
      false
      hasNumericResult
      true
      hasTaskVersionId
      false
      hasEvaluationDate
      true
      numericResultNullable
      value
      0.292
      standardDeviation
      0
      confidenceInterval
      0.0563619669903391
      hasStandardDeviation
      false
      hasConfidenceInterval
      true
      confidenceIntervalNullable
      0.0563619669903391
      hasUnevenConfidenceInterval
      false
      hasNumericResultPublic
      false
      customAdditionalResults
        hasNumericResultPrivate
        false
        task_id
        46
        sort_order
        DESCENDING
        source_name
        EM Marathi
        display_type
        percentages
        task_version
        1
        task_display_type
        PERCENTAGES
        normalization_status
        normalized
      published_at
      2025-07-15T11:49:31.790Z
      candidate_class
      modelVersion
      source_revision
      b32d007df5acbddafca23c7dd2beba8221c519cf7dd3445a1d7bc28740ffd0d8
      publisher_anchor
      openai/gpt-4o-2024-08-06
      metric_population
      51
      benchmark_evaluation
      true
      task_metadata_source
      benchmarkVersion
      candidate_metadata_source
      benchmarkVersionMappings
      evaluated at
      2025-06-16T00:00:00.000Z
      source model id
      GPT-4o
      confidence lower
      0.23563803300966088
      confidence upper
      0.34836196699033906
      Original result ↗

      Kaggle Benchmarks

      32.4%

      source id
      kaggleBenchmarks
      leaderboard name
      MultiLoKo German · EM German
      dimension
      reasoning
      metric
      MultiLoKo German · EM German
      rank
      14
      aggregation scope
      kaggle
      true
      status
      published
      license
      Apache-2.0
      direction
      higherIsBetter
      conditions
      task
      id
      37
      name
      EM German
      version
      1
      sort_order
      DESCENDING
      display_type
      percentages
      benchmark
      id
      60
      name
      MultiLoKo German
      type
      INDIVIDUAL
      owner
      metaresearch
      version
      1
      categories
      • multilingual and cross-lingual capabilities
      • general knowledge and reasoning
      version_id
      46
      maintainer_tier
      Kaggle-maintained
      maintenance_level
      KAGGLE_MAINTAINED
      candidate
      candidate_class
      modelVersion
      record_kind
      aggregate
      attribution
      Kaggle · metaresearch · MultiLoKo German
      license_url
      https://www.apache.org/licenses/LICENSE-2.0
      metric_kind
      MultiLoKo German · EM German
      rank_method
      competition rank among reported evaluations in identical conditions, including unlinked observations
      source_file
      46/leaderboard.json
      source_page
      https://www.kaggle.com/benchmarks/metaresearch/multiloko-german/versions/1
      measurements
      • name
        EM German
        unit
        percent
        scale
        1
        metric
        MultiLoKo German · EM German
        result
        resultCase
        numericResult
        booleanResult
        false
        numericResult
        value
        0.324
        standardDeviation
        0
        confidenceInterval
        0.0580128150237069
        hasStandardDeviation
        false
        hasConfidenceInterval
        true
        confidenceIntervalNullable
        0.0580128150237069
        hasUnevenConfidenceInterval
        false
        taskVersionId
        0
        evaluationDate
        2025-06-16T00:00:00Z
        hasBooleanResult
        false
        hasNumericResult
        true
        hasTaskVersionId
        false
        hasEvaluationDate
        true
        numericResultNullable
        value
        0.324
        standardDeviation
        0
        confidenceInterval
        0.0580128150237069
        hasStandardDeviation
        false
        hasConfidenceInterval
        true
        confidenceIntervalNullable
        0.0580128150237069
        hasUnevenConfidenceInterval
        false
        hasNumericResultPublic
        false
        customAdditionalResults
          hasNumericResultPrivate
          false
          task_id
          37
          sort_order
          DESCENDING
          source_name
          EM German
          display_type
          percentages
          task_version
          1
          task_display_type
          PERCENTAGES
          normalization_status
          normalized
        published_at
        2025-07-15T11:49:31.790Z
        candidate_class
        modelVersion
        source_revision
        b32d007df5acbddafca23c7dd2beba8221c519cf7dd3445a1d7bc28740ffd0d8
        publisher_anchor
        openai/gpt-4o-2024-08-06
        metric_population
        51
        benchmark_evaluation
        true
        task_metadata_source
        benchmarkVersion
        candidate_metadata_source
        benchmarkVersionMappings
        evaluated at
        2025-06-16T00:00:00.000Z
        source model id
        GPT-4o
        confidence lower
        0.2659871849762931
        confidence upper
        0.38201281502370693
        Original result ↗

        Kaggle Benchmarks

        26.8%

        source id
        kaggleBenchmarks
        leaderboard name
        MultiLoKo Japanese · EM Japanese
        dimension
        reasoning
        metric
        MultiLoKo Japanese · EM Japanese
        rank
        14
        aggregation scope
        kaggle
        true
        status
        published
        license
        Apache-2.0
        direction
        higherIsBetter
        conditions
        task
        id
        42
        name
        EM Japanese
        version
        1
        sort_order
        DESCENDING
        display_type
        percentages
        benchmark
        id
        65
        name
        MultiLoKo Japanese
        type
        INDIVIDUAL
        owner
        metaresearch
        version
        1
        categories
        • multilingual and cross-lingual capabilities
        • general knowledge and reasoning
        version_id
        51
        maintainer_tier
        Kaggle-maintained
        maintenance_level
        KAGGLE_MAINTAINED
        candidate
        candidate_class
        modelVersion
        record_kind
        aggregate
        attribution
        Kaggle · metaresearch · MultiLoKo Japanese
        license_url
        https://www.apache.org/licenses/LICENSE-2.0
        metric_kind
        MultiLoKo Japanese · EM Japanese
        rank_method
        competition rank among reported evaluations in identical conditions, including unlinked observations
        source_file
        51/leaderboard.json
        source_page
        https://www.kaggle.com/benchmarks/metaresearch/multiloko-japanese/versions/1
        measurements
        • name
          EM Japanese
          unit
          percent
          scale
          1
          metric
          MultiLoKo Japanese · EM Japanese
          result
          resultCase
          numericResult
          booleanResult
          false
          numericResult
          value
          0.268
          standardDeviation
          0
          confidenceInterval
          0.0549036255855108
          hasStandardDeviation
          false
          hasConfidenceInterval
          true
          confidenceIntervalNullable
          0.0549036255855108
          hasUnevenConfidenceInterval
          false
          taskVersionId
          0
          evaluationDate
          2025-06-16T00:00:00Z
          hasBooleanResult
          false
          hasNumericResult
          true
          hasTaskVersionId
          false
          hasEvaluationDate
          true
          numericResultNullable
          value
          0.268
          standardDeviation
          0
          confidenceInterval
          0.0549036255855108
          hasStandardDeviation
          false
          hasConfidenceInterval
          true
          confidenceIntervalNullable
          0.0549036255855108
          hasUnevenConfidenceInterval
          false
          hasNumericResultPublic
          false
          customAdditionalResults
            hasNumericResultPrivate
            false
            task_id
            42
            sort_order
            DESCENDING
            source_name
            EM Japanese
            display_type
            percentages
            task_version
            1
            task_display_type
            PERCENTAGES
            normalization_status
            normalized
          published_at
          2025-07-15T11:49:31.790Z
          candidate_class
          modelVersion
          source_revision
          b32d007df5acbddafca23c7dd2beba8221c519cf7dd3445a1d7bc28740ffd0d8
          publisher_anchor
          openai/gpt-4o-2024-08-06
          metric_population
          51
          benchmark_evaluation
          true
          task_metadata_source
          benchmarkVersion
          candidate_metadata_source
          benchmarkVersionMappings
          evaluated at
          2025-06-16T00:00:00.000Z
          source model id
          GPT-4o
          confidence lower
          0.21309637441448923
          confidence upper
          0.32290362558551083
          Original result ↗

          Kaggle Benchmarks

          38.4%

          source id
          kaggleBenchmarks
          leaderboard name
          MultiLoKo Traditional Mandarin · EM Traditional Mandarin
          dimension
          reasoning
          metric
          MultiLoKo Traditional Mandarin · EM Traditional Mandarin
          rank
          15
          aggregation scope
          kaggle
          true
          status
          published
          license
          Apache-2.0
          direction
          higherIsBetter
          conditions
          task
          id
          56
          name
          EM Traditional Mandarin
          version
          1
          sort_order
          DESCENDING
          display_type
          percentages
          benchmark
          id
          79
          name
          MultiLoKo Traditional Mandarin
          type
          INDIVIDUAL
          owner
          metaresearch
          version
          1
          categories
          • multilingual and cross-lingual capabilities
          • general knowledge and reasoning
          version_id
          65
          maintainer_tier
          Kaggle-maintained
          maintenance_level
          KAGGLE_MAINTAINED
          candidate
          candidate_class
          modelVersion
          record_kind
          aggregate
          attribution
          Kaggle · metaresearch · MultiLoKo Traditional Mandarin
          license_url
          https://www.apache.org/licenses/LICENSE-2.0
          metric_kind
          MultiLoKo Traditional Mandarin · EM Traditional Mandarin
          rank_method
          competition rank among reported evaluations in identical conditions, including unlinked observations
          source_file
          65/leaderboard.json
          source_page
          https://www.kaggle.com/benchmarks/metaresearch/multiloko-traditional-mandarin/versions/1
          measurements
          • name
            EM Traditional Mandarin
            unit
            percent
            scale
            1
            metric
            MultiLoKo Traditional Mandarin · EM Traditional Mandarin
            result
            resultCase
            numericResult
            booleanResult
            false
            numericResult
            value
            0.384
            standardDeviation
            0
            confidenceInterval
            0.0602884411900746
            hasStandardDeviation
            false
            hasConfidenceInterval
            true
            confidenceIntervalNullable
            0.0602884411900746
            hasUnevenConfidenceInterval
            false
            taskVersionId
            0
            evaluationDate
            2025-06-16T00:00:00Z
            hasBooleanResult
            false
            hasNumericResult
            true
            hasTaskVersionId
            false
            hasEvaluationDate
            true
            numericResultNullable
            value
            0.384
            standardDeviation
            0
            confidenceInterval
            0.0602884411900746
            hasStandardDeviation
            false
            hasConfidenceInterval
            true
            confidenceIntervalNullable
            0.0602884411900746
            hasUnevenConfidenceInterval
            false
            hasNumericResultPublic
            false
            customAdditionalResults
              hasNumericResultPrivate
              false
              task_id
              56
              sort_order
              DESCENDING
              source_name
              EM Traditional Mandarin
              display_type
              percentages
              task_version
              1
              task_display_type
              PERCENTAGES
              normalization_status
              normalized
            published_at
            2025-07-15T11:49:31.790Z
            candidate_class
            modelVersion
            source_revision
            b32d007df5acbddafca23c7dd2beba8221c519cf7dd3445a1d7bc28740ffd0d8
            publisher_anchor
            openai/gpt-4o-2024-08-06
            metric_population
            51
            benchmark_evaluation
            true
            task_metadata_source
            benchmarkVersion
            candidate_metadata_source
            benchmarkVersionMappings
            evaluated at
            2025-06-16T00:00:00.000Z
            source model id
            GPT-4o
            confidence lower
            0.3237115588099254
            confidence upper
            0.4442884411900746
            Original result ↗

            Kaggle Benchmarks

            35.6%

            source id
            kaggleBenchmarks
            leaderboard name
            MultiLoKo Simplified Mandarin · EM Simplified Mandarin
            dimension
            reasoning
            metric
            MultiLoKo Simplified Mandarin · EM Simplified Mandarin
            rank
            15
            aggregation scope
            kaggle
            true
            status
            published
            license
            Apache-2.0
            direction
            higherIsBetter
            conditions
            task
            id
            51
            name
            EM Simplified Mandarin
            version
            1
            sort_order
            DESCENDING
            display_type
            percentages
            benchmark
            id
            74
            name
            MultiLoKo Simplified Mandarin
            type
            INDIVIDUAL
            owner
            metaresearch
            version
            1
            categories
            • multilingual and cross-lingual capabilities
            • general knowledge and reasoning
            version_id
            60
            maintainer_tier
            Kaggle-maintained
            maintenance_level
            KAGGLE_MAINTAINED
            candidate
            candidate_class
            modelVersion
            record_kind
            aggregate
            attribution
            Kaggle · metaresearch · MultiLoKo Simplified Mandarin
            license_url
            https://www.apache.org/licenses/LICENSE-2.0
            metric_kind
            MultiLoKo Simplified Mandarin · EM Simplified Mandarin
            rank_method
            competition rank among reported evaluations in identical conditions, including unlinked observations
            source_file
            60/leaderboard.json
            source_page
            https://www.kaggle.com/benchmarks/metaresearch/multiloko-simplified-mandarin/versions/1
            measurements
            • name
              EM Simplified Mandarin
              unit
              percent
              scale
              1
              metric
              MultiLoKo Simplified Mandarin · EM Simplified Mandarin
              result
              resultCase
              numericResult
              booleanResult
              false
              numericResult
              value
              0.356
              standardDeviation
              0
              confidenceInterval
              0.0593534570203831
              hasStandardDeviation
              false
              hasConfidenceInterval
              true
              confidenceIntervalNullable
              0.0593534570203831
              hasUnevenConfidenceInterval
              false
              taskVersionId
              0
              evaluationDate
              2025-06-16T00:00:00Z
              hasBooleanResult
              false
              hasNumericResult
              true
              hasTaskVersionId
              false
              hasEvaluationDate
              true
              numericResultNullable
              value
              0.356
              standardDeviation
              0
              confidenceInterval
              0.0593534570203831
              hasStandardDeviation
              false
              hasConfidenceInterval
              true
              confidenceIntervalNullable
              0.0593534570203831
              hasUnevenConfidenceInterval
              false
              hasNumericResultPublic
              false
              customAdditionalResults
                hasNumericResultPrivate
                false
                task_id
                51
                sort_order
                DESCENDING
                source_name
                EM Simplified Mandarin
                display_type
                percentages
                task_version
                1
                task_display_type
                PERCENTAGES
                normalization_status
                normalized
              published_at
              2025-07-15T11:49:31.790Z
              candidate_class
              modelVersion
              source_revision
              b32d007df5acbddafca23c7dd2beba8221c519cf7dd3445a1d7bc28740ffd0d8
              publisher_anchor
              openai/gpt-4o-2024-08-06
              metric_population
              51
              benchmark_evaluation
              true
              task_metadata_source
              benchmarkVersion
              candidate_metadata_source
              benchmarkVersionMappings
              evaluated at
              2025-06-16T00:00:00.000Z
              source model id
              GPT-4o
              confidence lower
              0.2966465429796169
              confidence upper
              0.41535345702038307
              Original result ↗

              Kaggle Benchmarks

              31.6%

              source id
              kaggleBenchmarks
              leaderboard name
              MultiLoKo Russian · EM Russian
              dimension
              reasoning
              metric
              MultiLoKo Russian · EM Russian
              rank
              19
              aggregation scope
              kaggle
              true
              status
              published
              license
              Apache-2.0
              direction
              higherIsBetter
              conditions
              task
              id
              50
              name
              EM Russian
              version
              1
              sort_order
              DESCENDING
              display_type
              percentages
              benchmark
              id
              73
              name
              MultiLoKo Russian
              type
              INDIVIDUAL
              owner
              metaresearch
              version
              1
              categories
              • multilingual and cross-lingual capabilities
              • general knowledge and reasoning
              version_id
              59
              maintainer_tier
              Kaggle-maintained
              maintenance_level
              KAGGLE_MAINTAINED
              candidate
              candidate_class
              modelVersion
              record_kind
              aggregate
              attribution
              Kaggle · metaresearch · MultiLoKo Russian
              license_url
              https://www.apache.org/licenses/LICENSE-2.0
              metric_kind
              MultiLoKo Russian · EM Russian
              rank_method
              competition rank among reported evaluations in identical conditions, including unlinked observations
              source_file
              59/leaderboard.json
              source_page
              https://www.kaggle.com/benchmarks/metaresearch/multiloko-russian/versions/1
              measurements
              • name
                EM Russian
                unit
                percent
                scale
                1
                metric
                MultiLoKo Russian · EM Russian
                result
                resultCase
                numericResult
                booleanResult
                false
                numericResult
                value
                0.316
                standardDeviation
                0
                confidenceInterval
                0.0576301405634277
                hasStandardDeviation
                false
                hasConfidenceInterval
                true
                confidenceIntervalNullable
                0.0576301405634277
                hasUnevenConfidenceInterval
                false
                taskVersionId
                0
                evaluationDate
                2025-06-16T00:00:00Z
                hasBooleanResult
                false
                hasNumericResult
                true
                hasTaskVersionId
                false
                hasEvaluationDate
                true
                numericResultNullable
                value
                0.316
                standardDeviation
                0
                confidenceInterval
                0.0576301405634277
                hasStandardDeviation
                false
                hasConfidenceInterval
                true
                confidenceIntervalNullable
                0.0576301405634277
                hasUnevenConfidenceInterval
                false
                hasNumericResultPublic
                false
                customAdditionalResults
                  hasNumericResultPrivate
                  false
                  task_id
                  50
                  sort_order
                  DESCENDING
                  source_name
                  EM Russian
                  display_type
                  percentages
                  task_version
                  1
                  task_display_type
                  PERCENTAGES
                  normalization_status
                  normalized
                published_at
                2025-07-15T11:49:31.790Z
                candidate_class
                modelVersion
                source_revision
                b32d007df5acbddafca23c7dd2beba8221c519cf7dd3445a1d7bc28740ffd0d8
                publisher_anchor
                openai/gpt-4o-2024-08-06
                metric_population
                51
                benchmark_evaluation
                true
                task_metadata_source
                benchmarkVersion
                candidate_metadata_source
                benchmarkVersionMappings
                evaluated at
                2025-06-16T00:00:00.000Z
                source model id
                GPT-4o
                confidence lower
                0.2583698594365723
                confidence upper
                0.3736301405634277
                Original result ↗

                Kaggle Benchmarks

                11.6%

                source id
                kaggleBenchmarks
                leaderboard name
                MultiLoKo Khmer · EM Khmer
                dimension
                reasoning
                metric
                MultiLoKo Khmer · EM Khmer
                rank
                19
                aggregation scope
                kaggle
                true
                status
                published
                license
                Apache-2.0
                direction
                higherIsBetter
                conditions
                task
                id
                43
                name
                EM Khmer
                version
                1
                sort_order
                DESCENDING
                display_type
                percentages
                benchmark
                id
                66
                name
                MultiLoKo Khmer
                type
                INDIVIDUAL
                owner
                metaresearch
                version
                1
                categories
                • multilingual and cross-lingual capabilities
                • general knowledge and reasoning
                version_id
                52
                maintainer_tier
                Kaggle-maintained
                maintenance_level
                KAGGLE_MAINTAINED
                candidate
                candidate_class
                modelVersion
                record_kind
                aggregate
                attribution
                Kaggle · metaresearch · MultiLoKo Khmer
                license_url
                https://www.apache.org/licenses/LICENSE-2.0
                metric_kind
                MultiLoKo Khmer · EM Khmer
                rank_method
                competition rank among reported evaluations in identical conditions, including unlinked observations
                source_file
                52/leaderboard.json
                source_page
                https://www.kaggle.com/benchmarks/metaresearch/multiloko-khmer/versions/1
                measurements
                • name
                  EM Khmer
                  unit
                  percent
                  scale
                  1
                  metric
                  MultiLoKo Khmer · EM Khmer
                  result
                  resultCase
                  numericResult
                  booleanResult
                  false
                  numericResult
                  value
                  0.116
                  standardDeviation
                  0
                  confidenceInterval
                  0.0396947630454828
                  hasStandardDeviation
                  false
                  hasConfidenceInterval
                  true
                  confidenceIntervalNullable
                  0.0396947630454828
                  hasUnevenConfidenceInterval
                  false
                  taskVersionId
                  0
                  evaluationDate
                  2025-06-16T00:00:00Z
                  hasBooleanResult
                  false
                  hasNumericResult
                  true
                  hasTaskVersionId
                  false
                  hasEvaluationDate
                  true
                  numericResultNullable
                  value
                  0.116
                  standardDeviation
                  0
                  confidenceInterval
                  0.0396947630454828
                  hasStandardDeviation
                  false
                  hasConfidenceInterval
                  true
                  confidenceIntervalNullable
                  0.0396947630454828
                  hasUnevenConfidenceInterval
                  false
                  hasNumericResultPublic
                  false
                  customAdditionalResults
                    hasNumericResultPrivate
                    false
                    task_id
                    43
                    sort_order
                    DESCENDING
                    source_name
                    EM Khmer
                    display_type
                    percentages
                    task_version
                    1
                    task_display_type
                    PERCENTAGES
                    normalization_status
                    normalized
                  published_at
                  2025-07-15T11:49:31.790Z
                  candidate_class
                  modelVersion
                  source_revision
                  b32d007df5acbddafca23c7dd2beba8221c519cf7dd3445a1d7bc28740ffd0d8
                  publisher_anchor
                  openai/gpt-4o-2024-08-06
                  metric_population
                  51
                  benchmark_evaluation
                  true
                  task_metadata_source
                  benchmarkVersion
                  candidate_metadata_source
                  benchmarkVersionMappings
                  evaluated at
                  2025-06-16T00:00:00.000Z
                  source model id
                  GPT-4o
                  confidence lower
                  0.07630523695451721
                  confidence upper
                  0.1556947630454828
                  Original result ↗

                  Kaggle Benchmarks

                  18.4%

                  source id
                  kaggleBenchmarks
                  leaderboard name
                  MultiLoKo Cantonese · EM Cantonese
                  dimension
                  reasoning
                  metric
                  MultiLoKo Cantonese · EM Cantonese
                  rank
                  20
                  aggregation scope
                  kaggle
                  true
                  status
                  published
                  license
                  Apache-2.0
                  direction
                  higherIsBetter
                  conditions
                  task
                  id
                  32
                  name
                  EM Cantonese
                  version
                  1
                  sort_order
                  DESCENDING
                  display_type
                  percentages
                  benchmark
                  id
                  55
                  name
                  MultiLoKo Cantonese
                  type
                  INDIVIDUAL
                  owner
                  metaresearch
                  version
                  1
                  categories
                  • multilingual and cross-lingual capabilities
                  • general knowledge and reasoning
                  version_id
                  41
                  maintainer_tier
                  Kaggle-maintained
                  maintenance_level
                  KAGGLE_MAINTAINED
                  candidate
                  candidate_class
                  modelVersion
                  record_kind
                  aggregate
                  attribution
                  Kaggle · metaresearch · MultiLoKo Cantonese
                  license_url
                  https://www.apache.org/licenses/LICENSE-2.0
                  metric_kind
                  MultiLoKo Cantonese · EM Cantonese
                  rank_method
                  competition rank among reported evaluations in identical conditions, including unlinked observations
                  source_file
                  41/leaderboard.json
                  source_page
                  https://www.kaggle.com/benchmarks/metaresearch/multiloko-cantonese/versions/1
                  measurements
                  • name
                    EM Cantonese
                    unit
                    percent
                    scale
                    1
                    metric
                    MultiLoKo Cantonese · EM Cantonese
                    result
                    resultCase
                    numericResult
                    booleanResult
                    false
                    numericResult
                    value
                    0.184
                    standardDeviation
                    0
                    confidenceInterval
                    0.0480321556116024
                    hasStandardDeviation
                    false
                    hasConfidenceInterval
                    true
                    confidenceIntervalNullable
                    0.0480321556116024
                    hasUnevenConfidenceInterval
                    false
                    taskVersionId
                    0
                    evaluationDate
                    2025-06-16T00:00:00Z
                    hasBooleanResult
                    false
                    hasNumericResult
                    true
                    hasTaskVersionId
                    false
                    hasEvaluationDate
                    true
                    numericResultNullable
                    value
                    0.184
                    standardDeviation
                    0
                    confidenceInterval
                    0.0480321556116024
                    hasStandardDeviation
                    false
                    hasConfidenceInterval
                    true
                    confidenceIntervalNullable
                    0.0480321556116024
                    hasUnevenConfidenceInterval
                    false
                    hasNumericResultPublic
                    false
                    customAdditionalResults
                      hasNumericResultPrivate
                      false
                      task_id
                      32
                      sort_order
                      DESCENDING
                      source_name
                      EM Cantonese
                      display_type
                      percentages
                      task_version
                      1
                      task_display_type
                      PERCENTAGES
                      normalization_status
                      normalized
                    published_at
                    2025-07-15T11:49:31.790Z
                    candidate_class
                    modelVersion
                    source_revision
                    b32d007df5acbddafca23c7dd2beba8221c519cf7dd3445a1d7bc28740ffd0d8
                    publisher_anchor
                    openai/gpt-4o-2024-08-06
                    metric_population
                    51
                    benchmark_evaluation
                    true
                    task_metadata_source
                    benchmarkVersion
                    candidate_metadata_source
                    benchmarkVersionMappings
                    evaluated at
                    2025-06-16T00:00:00.000Z
                    source model id
                    GPT-4o
                    confidence lower
                    0.1359678443883976
                    confidence upper
                    0.2320321556116024
                    Original result ↗

                    OpenRouter Benchmarks

                    52.806%

                    source id
                    openRouterBenchmarks
                    leaderboard name
                    GPQA Diamond · Accuracy · OpenAI
                    dimension
                    knowledge
                    metric
                    GPQA Diamond · Accuracy
                    rank
                    23
                    sample size
                    594
                    aggregation scope
                    status
                    published
                    harness
                    url
                    https://github.com/OpenRouterTeam/benchmark-harness
                    name
                    OpenRouter benchmark harness
                    direction
                    higherIsBetter
                    run_count
                    3
                    conditions
                    window
                    newest 90 days; successful visible nonzero runs
                    dataset
                    benchmark-leaderboard
                    benchmark
                    id
                    gpqa_diamond
                    name
                    GPQA Diamond
                    record_kind
                    aggregate
                    provider_name
                    OpenAI
                    attribution
                    Source: OpenRouter (openrouter.ai), licensed under CC BY 4.0: reuse and republish with attribution to OpenRouter.
                    last_run_at
                    2026-10-02T23:16:52.382Z
                    license_url
                    https://creativecommons.org/licenses/by/4.0/
                    methodology
                    Rows aggregate the benchmark-runs dataset exactly as the leaderboard pages do, over each benchmark's non-zero-score runs from the newest 90 days: accuracy is the task-weighted mean of per-run accuracy, accuracy_std_dev is the population standard deviation of per-run accuracy (null with a single run), and the per-task averages divide summed cost, generation time, and output tokens by summed tasks. Rows are ordered by accuracy descending, then model name.
                    metric_kind
                    GPQA Diamond · Accuracy
                    rank_method
                    competition rank among reported evaluations in identical conditions, including unlinked observations
                    sample_size
                    594
                    source_file
                    benchmark-leaderboard.json
                    source_page
                    https://openrouter.ai/benchmarks/gpqa-diamond
                    measurements
                    accuracy
                    0.5280586666666666
                    accuracy_std_dev
                    0.004827144451490511
                    cost_per_task_usd
                    0.01799902861952862
                    output_tokens_per_task
                    1607.080808080808
                    latency_seconds_per_task
                    14.939786195286194
                    published_at
                    2026-11-16T01:46:02.653Z
                    snapshot_date
                    2026-11-16
                    source_revision
                    7ccaf17664252a3053dcb7df8ad0a46e72954a3b4524ddc1c8a7ef8a41b15568
                    transformations
                    TensorTrace preserves source results, converts latency to seconds and derives ranks within matching conditions. Individual runs have no derived rank.
                    metric_population
                    26
                    benchmark_evaluation
                    true
                    openrouter_permaslug
                    openai/gpt-4o-2024-08-06
                    source model id
                    openai/gpt-4o-2024-08-06
                    Original result ↗

                    Kaggle Benchmarks

                    0.711%

                    source id
                    kaggleBenchmarks
                    leaderboard name
                    BrowseComp · Score
                    dimension
                    knowledge
                    metric
                    BrowseComp · Score
                    rank
                    24
                    aggregation scope
                    kaggle
                    true
                    status
                    published
                    license
                    Apache-2.0
                    direction
                    higherIsBetter
                    conditions
                    task
                    id
                    26
                    name
                    Score
                    version
                    1
                    sort_order
                    DESCENDING
                    display_type
                    percentages
                    benchmark
                    id
                    49
                    name
                    BrowseComp
                    type
                    INDIVIDUAL
                    owner
                    openai
                    version
                    1
                    categories
                    • general knowledge and reasoning
                    • factuality and grounding
                    version_id
                    35
                    maintainer_tier
                    Kaggle-maintained
                    maintenance_level
                    KAGGLE_MAINTAINED
                    candidate
                    candidate_class
                    modelVersion
                    record_kind
                    aggregate
                    attribution
                    Kaggle · openai · BrowseComp
                    license_url
                    https://www.apache.org/licenses/LICENSE-2.0
                    metric_kind
                    BrowseComp · Score
                    rank_method
                    competition rank among reported evaluations in identical conditions, including unlinked observations
                    source_file
                    35/leaderboard.json
                    source_page
                    https://www.kaggle.com/benchmarks/openai/browsecomp/versions/1
                    measurements
                    • name
                      Score
                      unit
                      percent
                      scale
                      1
                      metric
                      BrowseComp · Score
                      result
                      resultCase
                      numericResult
                      booleanResult
                      false
                      numericResult
                      value
                      0.0071090047393364
                      standardDeviation
                      0
                      confidenceInterval
                      0.004627926213592
                      hasStandardDeviation
                      false
                      hasConfidenceInterval
                      true
                      confidenceIntervalNullable
                      0.004627926213592
                      hasUnevenConfidenceInterval
                      false
                      taskVersionId
                      0
                      evaluationDate
                      2025-06-13T00:00:00Z
                      hasBooleanResult
                      false
                      hasNumericResult
                      true
                      hasTaskVersionId
                      false
                      hasEvaluationDate
                      true
                      numericResultNullable
                      value
                      0.0071090047393364
                      standardDeviation
                      0
                      confidenceInterval
                      0.004627926213592
                      hasStandardDeviation
                      false
                      hasConfidenceInterval
                      true
                      confidenceIntervalNullable
                      0.004627926213592
                      hasUnevenConfidenceInterval
                      false
                      hasNumericResultPublic
                      false
                      customAdditionalResults
                        hasNumericResultPrivate
                        false
                        task_id
                        26
                        sort_order
                        DESCENDING
                        source_name
                        Score
                        display_type
                        percentages
                        task_version
                        1
                        task_display_type
                        PERCENTAGES
                        normalization_status
                        normalized
                      published_at
                      2025-07-15T14:25:30.827974100Z
                      candidate_class
                      modelVersion
                      source_revision
                      b32d007df5acbddafca23c7dd2beba8221c519cf7dd3445a1d7bc28740ffd0d8
                      publisher_anchor
                      openai/gpt-4o-2024-08-06
                      metric_population
                      42
                      benchmark_evaluation
                      true
                      task_metadata_source
                      benchmarkVersion
                      candidate_metadata_source
                      benchmarkVersionMappings
                      evaluated at
                      2025-06-13T00:00:00.000Z
                      source model id
                      GPT-4o
                      confidence lower
                      0.0024810785257443997
                      confidence upper
                      0.011736930952928399
                      Original result ↗

                      OpenRouter Benchmarks

                      43.111%

                      source id
                      openRouterBenchmarks
                      leaderboard name
                      tau2-bench Verified Airline · Accuracy · OpenAI
                      dimension
                      agentic
                      metric
                      tau2-bench Verified Airline · Accuracy
                      rank
                      25
                      sample size
                      150
                      aggregation scope
                      status
                      published
                      harness
                      url
                      https://github.com/OpenRouterTeam/benchmark-harness
                      name
                      OpenRouter benchmark harness
                      direction
                      higherIsBetter
                      run_count
                      3
                      conditions
                      window
                      newest 90 days; successful visible nonzero runs
                      dataset
                      benchmark-leaderboard
                      benchmark
                      id
                      tau_bench_verified_airline
                      name
                      tau2-bench Verified Airline
                      record_kind
                      aggregate
                      provider_name
                      OpenAI
                      attribution
                      Source: OpenRouter (openrouter.ai), licensed under CC BY 4.0: reuse and republish with attribution to OpenRouter.
                      last_run_at
                      2026-10-02T23:09:02.629Z
                      license_url
                      https://creativecommons.org/licenses/by/4.0/
                      methodology
                      Rows aggregate the benchmark-runs dataset exactly as the leaderboard pages do, over each benchmark's non-zero-score runs from the newest 90 days: accuracy is the task-weighted mean of per-run accuracy, accuracy_std_dev is the population standard deviation of per-run accuracy (null with a single run), and the per-task averages divide summed cost, generation time, and output tokens by summed tasks. Rows are ordered by accuracy descending, then model name.
                      metric_kind
                      tau2-bench Verified Airline · Accuracy
                      rank_method
                      competition rank among reported evaluations in identical conditions, including unlinked observations
                      sample_size
                      150
                      source_file
                      benchmark-leaderboard.json
                      source_page
                      https://openrouter.ai/benchmarks/tau2-bench-airline
                      measurements
                      accuracy
                      0.431111
                      accuracy_std_dev
                      0.00314238253559302
                      cost_per_task_usd
                      0.24174756666666666
                      output_tokens_per_task
                      2473.9733333333334
                      latency_seconds_per_task
                      46.57624
                      published_at
                      2026-11-16T01:46:02.653Z
                      snapshot_date
                      2026-11-16
                      source_revision
                      7ccaf17664252a3053dcb7df8ad0a46e72954a3b4524ddc1c8a7ef8a41b15568
                      transformations
                      TensorTrace preserves source results, converts latency to seconds and derives ranks within matching conditions. Individual runs have no derived rank.
                      metric_population
                      27
                      benchmark_evaluation
                      true
                      openrouter_permaslug
                      openai/gpt-4o-2024-08-06
                      source model id
                      openai/gpt-4o-2024-08-06
                      Original result ↗

                      Kaggle Benchmarks

                      0.077

                      source id
                      kaggleBenchmarks
                      leaderboard name
                      SciCode Main With Background · SciCode Main with Background
                      dimension
                      coding
                      metric
                      SciCode Main With Background · SciCode Main with Background
                      rank
                      33
                      aggregation scope
                      kaggle
                      true
                      status
                      published
                      license
                      Apache-2.0
                      direction
                      higherIsBetter
                      conditions
                      task
                      id
                      40201
                      name
                      SciCode Main with Background
                      version
                      3
                      sort_order
                      DESCENDING
                      display_type
                      rawValues
                      benchmark
                      id
                      13
                      name
                      SciCode Main With Background
                      type
                      INDIVIDUAL
                      owner
                      open-benchmarks
                      version
                      1
                      categories
                      • general knowledge and reasoning
                      • coding
                      • science
                      version_id
                      13
                      maintainer_tier
                      Kaggle-maintained
                      maintenance_level
                      KAGGLE_MAINTAINED
                      candidate
                      candidate_class
                      modelVersion
                      record_kind
                      aggregate
                      attribution
                      Kaggle · open-benchmarks · SciCode Main With Background
                      license_url
                      https://www.apache.org/licenses/LICENSE-2.0
                      metric_kind
                      SciCode Main With Background · SciCode Main with Background
                      rank_method
                      competition rank among reported evaluations in identical conditions, including unlinked observations
                      source_file
                      13/leaderboard.json
                      source_page
                      https://www.kaggle.com/benchmarks/open-benchmarks/scicode-main-with-background/versions/1
                      measurements
                      • name
                        SciCode Main with Background
                        unit
                        score
                        scale
                        1
                        metric
                        SciCode Main With Background · SciCode Main with Background
                        result
                        resultCase
                        numericResult
                        booleanResult
                        false
                        numericResult
                        value
                        0.0769230769230769
                        standardDeviation
                        0
                        confidenceInterval
                        0.0647796628574181
                        hasStandardDeviation
                        false
                        hasConfidenceInterval
                        true
                        confidenceIntervalNullable
                        0.0647796628574181
                        hasUnevenConfidenceInterval
                        false
                        taskVersionId
                        0
                        evaluationDate
                        2025-05-22T00:00:00Z
                        hasBooleanResult
                        false
                        hasNumericResult
                        true
                        hasTaskVersionId
                        false
                        hasEvaluationDate
                        true
                        numericResultNullable
                        value
                        0.0769230769230769
                        standardDeviation
                        0
                        confidenceInterval
                        0.0647796628574181
                        hasStandardDeviation
                        false
                        hasConfidenceInterval
                        true
                        confidenceIntervalNullable
                        0.0647796628574181
                        hasUnevenConfidenceInterval
                        false
                        hasNumericResultPublic
                        false
                        customAdditionalResults
                          hasNumericResultPrivate
                          false
                          task_id
                          40201
                          task_url
                          /benchmarks/tasks/benchmarkler/scicode-background-scicode-main-with-background
                          sort_order
                          DESCENDING
                          source_name
                          SciCode Main with Background
                          display_type
                          rawValues
                          task_version
                          3
                          task_display_type
                          RAW_VALUES
                          normalization_status
                          normalized
                        published_at
                        2025-07-15T11:49:31.790Z
                        candidate_class
                        modelVersion
                        source_revision
                        b32d007df5acbddafca23c7dd2beba8221c519cf7dd3445a1d7bc28740ffd0d8
                        publisher_anchor
                        openai/gpt-4o-2024-08-06
                        metric_population
                        57
                        benchmark_evaluation
                        true
                        task_metadata_source
                        benchmarkVersion
                        candidate_metadata_source
                        benchmarkVersionMappings
                        evaluated at
                        2025-05-22T00:00:00.000Z
                        source model id
                        GPT-4o
                        confidence lower
                        0.012143414065658803
                        confidence upper
                        0.141702739780495
                        Original result ↗

                        Kaggle Benchmarks

                        1.538%

                        source id
                        kaggleBenchmarks
                        leaderboard name
                        SciCode Main Standard · SciCode Main Standard
                        dimension
                        coding
                        metric
                        SciCode Main Standard · SciCode Main Standard
                        rank
                        36
                        aggregation scope
                        kaggle
                        true
                        status
                        published
                        license
                        Apache-2.0
                        direction
                        higherIsBetter
                        conditions
                        task
                        id
                        40105
                        name
                        SciCode Main Standard
                        version
                        3
                        sort_order
                        DESCENDING
                        display_type
                        percentages
                        benchmark
                        id
                        12
                        name
                        SciCode Main Standard
                        type
                        INDIVIDUAL
                        owner
                        open-benchmarks
                        version
                        1
                        categories
                        • general knowledge and reasoning
                        • coding
                        • science
                        version_id
                        12
                        maintainer_tier
                        Kaggle-maintained
                        maintenance_level
                        KAGGLE_MAINTAINED
                        candidate
                        candidate_class
                        modelVersion
                        record_kind
                        aggregate
                        attribution
                        Kaggle · open-benchmarks · SciCode Main Standard
                        license_url
                        https://www.apache.org/licenses/LICENSE-2.0
                        metric_kind
                        SciCode Main Standard · SciCode Main Standard
                        rank_method
                        competition rank among reported evaluations in identical conditions, including unlinked observations
                        source_file
                        12/leaderboard.json
                        source_page
                        https://www.kaggle.com/benchmarks/open-benchmarks/scicode-main-standard/versions/1
                        measurements
                        • name
                          SciCode Main Standard
                          unit
                          percent
                          scale
                          1
                          metric
                          SciCode Main Standard · SciCode Main Standard
                          result
                          resultCase
                          numericResult
                          booleanResult
                          false
                          numericResult
                          value
                          0.0153846153846153
                          standardDeviation
                          0
                          confidenceInterval
                          0.0299204446309948
                          hasStandardDeviation
                          false
                          hasConfidenceInterval
                          true
                          confidenceIntervalNullable
                          0.0299204446309948
                          hasUnevenConfidenceInterval
                          false
                          taskVersionId
                          0
                          evaluationDate
                          2025-05-22T00:00:00Z
                          hasBooleanResult
                          false
                          hasNumericResult
                          true
                          hasTaskVersionId
                          false
                          hasEvaluationDate
                          true
                          numericResultNullable
                          value
                          0.0153846153846153
                          standardDeviation
                          0
                          confidenceInterval
                          0.0299204446309948
                          hasStandardDeviation
                          false
                          hasConfidenceInterval
                          true
                          confidenceIntervalNullable
                          0.0299204446309948
                          hasUnevenConfidenceInterval
                          false
                          hasNumericResultPublic
                          false
                          customAdditionalResults
                            hasNumericResultPrivate
                            false
                            task_id
                            40105
                            task_url
                            /benchmarks/tasks/benchmarkler/scicode-standard-scicode-main-standard
                            sort_order
                            DESCENDING
                            source_name
                            SciCode Main Standard
                            display_type
                            percentages
                            task_version
                            3
                            task_display_type
                            PERCENTAGES
                            normalization_status
                            normalized
                          published_at
                          2025-07-15T11:49:31.790Z
                          candidate_class
                          modelVersion
                          source_revision
                          b32d007df5acbddafca23c7dd2beba8221c519cf7dd3445a1d7bc28740ffd0d8
                          publisher_anchor
                          openai/gpt-4o-2024-08-06
                          metric_population
                          57
                          benchmark_evaluation
                          true
                          task_metadata_source
                          benchmarkVersion
                          candidate_metadata_source
                          benchmarkVersionMappings
                          evaluated at
                          2025-05-22T00:00:00.000Z
                          source model id
                          GPT-4o
                          confidence lower
                          -0.014535829246379499
                          confidence upper
                          0.0453050600156101
                          Original result ↗

                          SWE-bench

                          36%

                          source id
                          sweBench
                          leaderboard name
                          SWE-bench Lite
                          dimension
                          coding
                          metric
                          Resolved tasks
                          rank
                          37
                          evaluated at
                          2024-11-13T00:00:00.000Z
                          Original result ↗

                          Kaggle Benchmarks

                          7.692%

                          source id
                          kaggleBenchmarks
                          leaderboard name
                          SciCode · SciCode Main With Background
                          dimension
                          coding
                          metric
                          SciCode · SciCode Main With Background
                          rank
                          37
                          aggregation scope
                          kaggle
                          true
                          status
                          published
                          license
                          Apache-2.0
                          direction
                          higherIsBetter
                          conditions
                          task
                          id
                          40201
                          name
                          SciCode Main With Background
                          version
                          3
                          sort_order
                          DESCENDING
                          display_type
                          percentages
                          benchmark
                          id
                          16
                          name
                          SciCode
                          type
                          SUITE
                          owner
                          open-benchmarks
                          version
                          1
                          categories
                          • general knowledge and reasoning
                          • coding
                          • science
                          version_id
                          16
                          maintainer_tier
                          Kaggle-maintained
                          maintenance_level
                          KAGGLE_MAINTAINED
                          candidate
                          candidate_class
                          modelVersion
                          record_kind
                          aggregate
                          attribution
                          Kaggle · open-benchmarks · SciCode
                          license_url
                          https://www.apache.org/licenses/LICENSE-2.0
                          metric_kind
                          SciCode · SciCode Main With Background
                          rank_method
                          competition rank among reported evaluations in identical conditions, including unlinked observations
                          source_file
                          16/leaderboard.json
                          source_page
                          https://www.kaggle.com/benchmarks/open-benchmarks/scicode/versions/1
                          measurements
                          • name
                            SciCode Main With Background
                            unit
                            percent
                            scale
                            1
                            metric
                            SciCode · SciCode Main With Background
                            result
                            resultCase
                            numericResult
                            booleanResult
                            false
                            numericResult
                            value
                            0.0769230769230769
                            standardDeviation
                            0
                            confidenceInterval
                            0.0647796628574181
                            hasStandardDeviation
                            false
                            hasConfidenceInterval
                            true
                            confidenceIntervalNullable
                            0.0647796628574181
                            hasUnevenConfidenceInterval
                            false
                            taskVersionId
                            0
                            evaluationDate
                            2025-05-22T00:00:00Z
                            hasBooleanResult
                            false
                            hasNumericResult
                            true
                            hasTaskVersionId
                            false
                            hasEvaluationDate
                            true
                            numericResultNullable
                            value
                            0.0769230769230769
                            standardDeviation
                            0
                            confidenceInterval
                            0.0647796628574181
                            hasStandardDeviation
                            false
                            hasConfidenceInterval
                            true
                            confidenceIntervalNullable
                            0.0647796628574181
                            hasUnevenConfidenceInterval
                            false
                            hasNumericResultPublic
                            false
                            customAdditionalResults
                              hasNumericResultPrivate
                              false
                              task_id
                              40201
                              task_url
                              /benchmarks/open-benchmarks/scicode-main-with-background/versions/1
                              sort_order
                              DESCENDING
                              source_name
                              SciCode Main With Background
                              display_type
                              percentages
                              task_version
                              3
                              task_display_type
                              RAW_VALUES
                              normalization_status
                              normalized
                            published_at
                            2025-10-21T20:29:26.251536900Z
                            candidate_class
                            modelVersion
                            source_revision
                            b32d007df5acbddafca23c7dd2beba8221c519cf7dd3445a1d7bc28740ffd0d8
                            publisher_anchor
                            openai/gpt-4o-2024-08-06
                            metric_population
                            62
                            benchmark_evaluation
                            true
                            task_metadata_source
                            benchmarkVersion
                            candidate_metadata_source
                            benchmarkVersionMappings
                            evaluated at
                            2025-05-22T00:00:00.000Z
                            source model id
                            GPT-4o
                            confidence lower
                            0.012143414065658803
                            confidence upper
                            0.141702739780495
                            Original result ↗

                            Kaggle Benchmarks

                            88.086%

                            source id
                            kaggleBenchmarks
                            leaderboard name
                            MMLU · mmlu_task
                            dimension
                            reasoning
                            metric
                            MMLU · mmlu_task
                            rank
                            39
                            aggregation scope
                            kaggle
                            true
                            status
                            published
                            license
                            Apache-2.0
                            direction
                            higherIsBetter
                            conditions
                            task
                            id
                            40139
                            name
                            mmlu_task
                            version
                            7
                            sort_order
                            DESCENDING
                            display_type
                            percentages
                            benchmark
                            id
                            42
                            name
                            MMLU
                            type
                            INDIVIDUAL
                            owner
                            open-benchmarks
                            version
                            1
                            categories
                            • general knowledge and reasoning
                            version_id
                            28
                            maintainer_tier
                            Kaggle-maintained
                            maintenance_level
                            KAGGLE_MAINTAINED
                            candidate
                            candidate_class
                            modelVersion
                            record_kind
                            aggregate
                            attribution
                            Kaggle · open-benchmarks · MMLU
                            license_url
                            https://www.apache.org/licenses/LICENSE-2.0
                            metric_kind
                            MMLU · mmlu_task
                            rank_method
                            competition rank among reported evaluations in identical conditions, including unlinked observations
                            source_file
                            28/leaderboard.json
                            source_page
                            https://www.kaggle.com/benchmarks/open-benchmarks/mmlu/versions/1
                            measurements
                            • name
                              mmlu_task
                              unit
                              percent
                              scale
                              1
                              metric
                              MMLU · mmlu_task
                              result
                              resultCase
                              numericResult
                              booleanResult
                              false
                              numericResult
                              value
                              0.8808574277168495
                              standardDeviation
                              0
                              confidenceInterval
                              0.0053582121301719
                              hasStandardDeviation
                              false
                              hasConfidenceInterval
                              true
                              confidenceIntervalNullable
                              0.0053582121301719
                              hasUnevenConfidenceInterval
                              false
                              taskVersionId
                              0
                              evaluationDate
                              2025-05-29T00:00:00Z
                              hasBooleanResult
                              false
                              hasNumericResult
                              true
                              hasTaskVersionId
                              false
                              hasEvaluationDate
                              true
                              numericResultNullable
                              value
                              0.8808574277168495
                              standardDeviation
                              0
                              confidenceInterval
                              0.0053582121301719
                              hasStandardDeviation
                              false
                              hasConfidenceInterval
                              true
                              confidenceIntervalNullable
                              0.0053582121301719
                              hasUnevenConfidenceInterval
                              false
                              hasNumericResultPublic
                              false
                              customAdditionalResults
                                hasNumericResultPrivate
                                false
                                task_id
                                40139
                                task_url
                                /benchmarks/tasks/benchmarkler/mmlu-task
                                sort_order
                                DESCENDING
                                source_name
                                mmlu_task
                                display_type
                                percentages
                                task_version
                                7
                                task_display_type
                                PERCENTAGES
                                normalization_status
                                normalized
                              published_at
                              2025-07-15T11:49:31.790Z
                              candidate_class
                              modelVersion
                              source_revision
                              b32d007df5acbddafca23c7dd2beba8221c519cf7dd3445a1d7bc28740ffd0d8
                              publisher_anchor
                              openai/gpt-4o-2024-08-06
                              metric_population
                              61
                              benchmark_evaluation
                              true
                              task_metadata_source
                              benchmarkVersion
                              candidate_metadata_source
                              benchmarkVersionMappings
                              evaluated at
                              2025-05-29T00:00:00.000Z
                              source model id
                              GPT-4o
                              confidence lower
                              0.8754992155866776
                              confidence upper
                              0.8862156398470213
                              Original result ↗

                              Kaggle Benchmarks

                              0.337

                              source id
                              kaggleBenchmarks
                              leaderboard name
                              SciCode Subproblem With Background · SciCode Subproblem with Background
                              dimension
                              coding
                              metric
                              SciCode Subproblem With Background · SciCode Subproblem with Background
                              rank
                              39
                              aggregation scope
                              kaggle
                              true
                              status
                              published
                              license
                              Apache-2.0
                              direction
                              higherIsBetter
                              conditions
                              task
                              id
                              40200
                              name
                              SciCode Subproblem with Background
                              version
                              3
                              sort_order
                              DESCENDING
                              display_type
                              rawValues
                              benchmark
                              id
                              15
                              name
                              SciCode Subproblem With Background
                              type
                              INDIVIDUAL
                              owner
                              open-benchmarks
                              version
                              1
                              categories
                              • general knowledge and reasoning
                              • coding
                              • science
                              version_id
                              15
                              maintainer_tier
                              Kaggle-maintained
                              maintenance_level
                              KAGGLE_MAINTAINED
                              candidate
                              candidate_class
                              modelVersion
                              record_kind
                              aggregate
                              attribution
                              Kaggle · open-benchmarks · SciCode Subproblem With Background
                              license_url
                              https://www.apache.org/licenses/LICENSE-2.0
                              metric_kind
                              SciCode Subproblem With Background · SciCode Subproblem with Background
                              rank_method
                              competition rank among reported evaluations in identical conditions, including unlinked observations
                              source_file
                              15/leaderboard.json
                              source_page
                              https://www.kaggle.com/benchmarks/open-benchmarks/scicode-subproblem-with-background/versions/1
                              measurements
                              • name
                                SciCode Subproblem with Background
                                unit
                                score
                                scale
                                1
                                metric
                                SciCode Subproblem With Background · SciCode Subproblem with Background
                                result
                                resultCase
                                numericResult
                                booleanResult
                                false
                                numericResult
                                value
                                0.3367697594501718
                                standardDeviation
                                0
                                confidenceInterval
                                0.0543000588021355
                                hasStandardDeviation
                                false
                                hasConfidenceInterval
                                true
                                confidenceIntervalNullable
                                0.0543000588021355
                                hasUnevenConfidenceInterval
                                false
                                taskVersionId
                                0
                                evaluationDate
                                2025-05-22T00:00:00Z
                                hasBooleanResult
                                false
                                hasNumericResult
                                true
                                hasTaskVersionId
                                false
                                hasEvaluationDate
                                true
                                numericResultNullable
                                value
                                0.3367697594501718
                                standardDeviation
                                0
                                confidenceInterval
                                0.0543000588021355
                                hasStandardDeviation
                                false
                                hasConfidenceInterval
                                true
                                confidenceIntervalNullable
                                0.0543000588021355
                                hasUnevenConfidenceInterval
                                false
                                hasNumericResultPublic
                                false
                                customAdditionalResults
                                  hasNumericResultPrivate
                                  false
                                  task_id
                                  40200
                                  task_url
                                  /benchmarks/tasks/benchmarkler/scicode-background-scicode-subproblem-with-background
                                  sort_order
                                  DESCENDING
                                  source_name
                                  SciCode Subproblem with Background
                                  display_type
                                  rawValues
                                  task_version
                                  3
                                  task_display_type
                                  RAW_VALUES
                                  normalization_status
                                  normalized
                                published_at
                                2025-07-15T11:49:31.790Z
                                candidate_class
                                modelVersion
                                source_revision
                                b32d007df5acbddafca23c7dd2beba8221c519cf7dd3445a1d7bc28740ffd0d8
                                publisher_anchor
                                openai/gpt-4o-2024-08-06
                                metric_population
                                57
                                benchmark_evaluation
                                true
                                task_metadata_source
                                benchmarkVersion
                                candidate_metadata_source
                                benchmarkVersionMappings
                                evaluated at
                                2025-05-22T00:00:00.000Z
                                source model id
                                GPT-4o
                                confidence lower
                                0.2824697006480363
                                confidence upper
                                0.3910698182523073
                                Original result ↗

                                Kaggle Benchmarks

                                0.241

                                source id
                                kaggleBenchmarks
                                leaderboard name
                                SciCode Subproblem Standard · SciCode Subproblem Standard
                                dimension
                                coding
                                metric
                                SciCode Subproblem Standard · SciCode Subproblem Standard
                                rank
                                39
                                aggregation scope
                                kaggle
                                true
                                status
                                published
                                license
                                Apache-2.0
                                direction
                                higherIsBetter
                                conditions
                                task
                                id
                                40106
                                name
                                SciCode Subproblem Standard
                                version
                                3
                                sort_order
                                DESCENDING
                                display_type
                                rawValues
                                benchmark
                                id
                                14
                                name
                                SciCode Subproblem Standard
                                type
                                INDIVIDUAL
                                owner
                                open-benchmarks
                                version
                                1
                                categories
                                • general knowledge and reasoning
                                • coding
                                • science
                                version_id
                                14
                                maintainer_tier
                                Kaggle-maintained
                                maintenance_level
                                KAGGLE_MAINTAINED
                                candidate
                                candidate_class
                                modelVersion
                                record_kind
                                aggregate
                                attribution
                                Kaggle · open-benchmarks · SciCode Subproblem Standard
                                license_url
                                https://www.apache.org/licenses/LICENSE-2.0
                                metric_kind
                                SciCode Subproblem Standard · SciCode Subproblem Standard
                                rank_method
                                competition rank among reported evaluations in identical conditions, including unlinked observations
                                source_file
                                14/leaderboard.json
                                source_page
                                https://www.kaggle.com/benchmarks/open-benchmarks/scicode-subproblem-standard/versions/1
                                measurements
                                • name
                                  SciCode Subproblem Standard
                                  unit
                                  score
                                  scale
                                  1
                                  metric
                                  SciCode Subproblem Standard · SciCode Subproblem Standard
                                  result
                                  resultCase
                                  numericResult
                                  booleanResult
                                  false
                                  numericResult
                                  value
                                  0.2405498281786941
                                  standardDeviation
                                  0
                                  confidenceInterval
                                  0.0491081715605116
                                  hasStandardDeviation
                                  false
                                  hasConfidenceInterval
                                  true
                                  confidenceIntervalNullable
                                  0.0491081715605116
                                  hasUnevenConfidenceInterval
                                  false
                                  taskVersionId
                                  0
                                  evaluationDate
                                  2025-05-22T00:00:00Z
                                  hasBooleanResult
                                  false
                                  hasNumericResult
                                  true
                                  hasTaskVersionId
                                  false
                                  hasEvaluationDate
                                  true
                                  numericResultNullable
                                  value
                                  0.2405498281786941
                                  standardDeviation
                                  0
                                  confidenceInterval
                                  0.0491081715605116
                                  hasStandardDeviation
                                  false
                                  hasConfidenceInterval
                                  true
                                  confidenceIntervalNullable
                                  0.0491081715605116
                                  hasUnevenConfidenceInterval
                                  false
                                  hasNumericResultPublic
                                  false
                                  customAdditionalResults
                                    hasNumericResultPrivate
                                    false
                                    task_id
                                    40106
                                    task_url
                                    /benchmarks/tasks/benchmarkler/scicode-standard-scicode-subproblem-standard
                                    sort_order
                                    DESCENDING
                                    source_name
                                    SciCode Subproblem Standard
                                    display_type
                                    rawValues
                                    task_version
                                    3
                                    task_display_type
                                    RAW_VALUES
                                    normalization_status
                                    normalized
                                  published_at
                                  2025-07-15T11:49:31.790Z
                                  candidate_class
                                  modelVersion
                                  source_revision
                                  b32d007df5acbddafca23c7dd2beba8221c519cf7dd3445a1d7bc28740ffd0d8
                                  publisher_anchor
                                  openai/gpt-4o-2024-08-06
                                  metric_population
                                  57
                                  benchmark_evaluation
                                  true
                                  task_metadata_source
                                  benchmarkVersion
                                  candidate_metadata_source
                                  benchmarkVersionMappings
                                  evaluated at
                                  2025-05-22T00:00:00.000Z
                                  source model id
                                  GPT-4o
                                  confidence lower
                                  0.19144165661818252
                                  confidence upper
                                  0.2896579997392057
                                  Original result ↗

                                  OpenRouter Benchmarks

                                  51.515%

                                  source id
                                  openRouterBenchmarks
                                  leaderboard name
                                  GPQA Diamond · Accuracy · Azure
                                  dimension
                                  knowledge
                                  metric
                                  GPQA Diamond · Accuracy
                                  rank
                                  41
                                  sample size
                                  396
                                  aggregation scope
                                  status
                                  published
                                  harness
                                  url
                                  https://github.com/OpenRouterTeam/benchmark-harness
                                  name
                                  OpenRouter benchmark harness
                                  direction
                                  higherIsBetter
                                  run_count
                                  2
                                  conditions
                                  window
                                  newest 90 days; successful visible nonzero runs
                                  dataset
                                  benchmark-leaderboard
                                  benchmark
                                  id
                                  gpqa_diamond
                                  name
                                  GPQA Diamond
                                  record_kind
                                  aggregate
                                  provider_name
                                  Azure
                                  attribution
                                  Source: OpenRouter (openrouter.ai), licensed under CC BY 4.0: reuse and republish with attribution to OpenRouter.
                                  last_run_at
                                  2026-10-01T16:26:08.836Z
                                  license_url
                                  https://creativecommons.org/licenses/by/4.0/
                                  methodology
                                  Rows aggregate the benchmark-runs dataset exactly as the leaderboard pages do, over each benchmark's non-zero-score runs from the newest 90 days: accuracy is the task-weighted mean of per-run accuracy, accuracy_std_dev is the population standard deviation of per-run accuracy (null with a single run), and the per-task averages divide summed cost, generation time, and output tokens by summed tasks. Rows are ordered by accuracy descending, then model name.
                                  metric_kind
                                  GPQA Diamond · Accuracy
                                  rank_method
                                  competition rank among reported evaluations in identical conditions, including unlinked observations
                                  sample_size
                                  396
                                  source_file
                                  benchmark-leaderboard.json
                                  source_page
                                  https://openrouter.ai/benchmarks/gpqa-diamond
                                  measurements
                                  accuracy
                                  0.5151519999999999
                                  accuracy_std_dev
                                  0.006733999999999962
                                  cost_per_task_usd
                                  0.018041959595959597
                                  output_tokens_per_task
                                  1611.5353535353536
                                  latency_seconds_per_task
                                  13.4984797979798
                                  published_at
                                  2026-11-16T01:46:02.653Z
                                  snapshot_date
                                  2026-11-16
                                  source_revision
                                  7ccaf17664252a3053dcb7df8ad0a46e72954a3b4524ddc1c8a7ef8a41b15568
                                  transformations
                                  TensorTrace preserves source results, converts latency to seconds and derives ranks within matching conditions. Individual runs have no derived rank.
                                  metric_population
                                  43
                                  benchmark_evaluation
                                  true
                                  openrouter_permaslug
                                  openai/gpt-4o-2024-08-06
                                  source model id
                                  openai/gpt-4o-2024-08-06
                                  Original result ↗

                                  Published prices

                                  1–3 / 3

                                  openai/gpt-4o-2024-08-06 ↗

                                  $1.25

                                  Inference provider
                                  openRouter
                                  Platform
                                  OpenRouter
                                  Unit
                                  cache_read_tokens_per_million
                                  Currency
                                  USD
                                  Billing conditions

                                  Safety assessments

                                  1–1 / 1

                                  Foundation Model Transparency Index 2025 ↗

                                  35 / 100

                                  source id
                                  stanfordFMTI
                                  canonical model id
                                  openai/gpt-4o-2024-08-06
                                  taxonomy
                                  transparency
                                  hazard category
                                  transparency
                                  hazard name
                                  Organization transparency
                                  policy area
                                  OpenAI
                                  assessment type
                                  organization-level disclosure
                                  metric
                                  disclosure_score
                                  result label
                                  35 / 100
                                  sample size
                                  100
                                  updated at
                                  2026-11-09T18:08:05.227Z

                                  Source specification observations

                                  1–3 / 3

                                  openRouter ↗

                                  field
                                  context_length
                                  value
                                  128000
                                  normalized value
                                  128000
                                  scope
                                  source-model
                                  source id
                                  openRouter
                                  confidence
                                  1

                                  openRouter ↗

                                  field
                                  knowledge_cutoff
                                  reported field
                                  knowledge_cutoff
                                  value
                                  2023-10-31
                                  normalized value
                                  2023-10-31
                                  precision
                                  day
                                  scope
                                  source-model
                                  source id
                                  openRouter
                                  confidence
                                  1

                                  openRouter ↗

                                  field
                                  max_output_tokens
                                  value
                                  16384
                                  normalized value
                                  16384
                                  scope
                                  source-model
                                  source id
                                  openRouter
                                  confidence
                                  1

                                  Original sources

                                  1–7 / 7

                                  OpenRouter · Model page ↗

                                  GitHub · Catalog ↗

                                  GitHub · Catalog ↗

                                  Kaggle · Catalog ↗

                                  OpenRouter · Catalog ↗

                                  OpenRouter · Catalog ↗

                                  www.swebench.com · Catalog ↗

                                  Source links

                                  1–7 / 7

                                  Explore more models

                                  Models

                                  Loading
                                  EVIDENCE / COMPARISON

                                  Benchmarks

                                  Explore every benchmark

                                  Loading

                                  These charts contain model-linked results. The charts above also include unlinked source profiles. Compare scores within each evaluation. Model quality, usage, transparency and system performance measure different things.

                                  Install TensorTrace