{
  "schemaVersion": 1,
  "benchmark": {
    "id": "vals-legal-research",
    "name": "Legal Research Bench · Vals AI",
    "version": "1 · private test set",
    "category": "work",
    "question": "Can an agent do legal research it can cite?",
    "summary": "Case and statute research across eight areas of US law, scored against citation-backed answers.",
    "source": {
      "name": "Vals AI",
      "url": "https://www.vals.ai/benchmarks/legal_research",
      "methodologyUrl": "https://www.vals.ai/about"
    },
    "measure": "Accuracy on research questions across administrative, business, civil, constitutional, criminal, family, health, and immigration law.",
    "comparisonRule": "Compare within Legal Research Bench version 1. Per-area scores are components, not separate leaderboards.",
    "limitations": [
      "The test set is private and was built with partner law firms, so no independent party can reproduce these scores.",
      "US law only. The result says nothing about research in another jurisdiction.",
      "A citation-backed answer that scores well is not legal advice and has not been checked by a lawyer for a specific matter.",
      "Model names are the identifiers Vals publishes, not aicharts canonical names."
    ],
    "coverage": "charted",
    "tags": [
      "legal",
      "research",
      "citations",
      "case law",
      "agentic",
      "professional work",
      "Vals AI",
      "independent evaluator"
    ]
  },
  "dataset": {
    "benchmarkId": "vals-legal-research",
    "version": "1 · private test set",
    "score": {
      "label": "Accuracy",
      "unit": "%",
      "direction": "higher",
      "minimum": 0,
      "maximum": 100
    },
    "source": {
      "name": "Vals AI · Legal Research Bench",
      "url": "https://www.vals.ai/benchmarks/legal_research",
      "retrievedAt": "2026-09-28T18:16:23.422Z",
      "revision": "legal_research v1"
    },
    "observedAt": "2026-09-27T00:00:00.000Z",
    "evidenceLabel": "Independent evaluator · private test set",
    "configurationLabel": "Model at the effort Vals selected",
    "comparabilityNote": "Private evaluation run by Vals with partner law firms. Dollar values are the publisher's average cost per test. Model names are the identifiers Vals publishes, not aicharts canonical names.",
    "costLabel": "Average cost per test (USD)",
    "points": [
      {
        "id": "alibaba-qwen3-6-plus",
        "label": "alibaba/qwen3.6-plus",
        "model": "alibaba/qwen3.6-plus",
        "provider": "Alibaba",
        "harness": null,
        "effort": null,
        "score": 14.904,
        "costUsd": 0.697997,
        "uncertainty": {
          "lower": 12.429,
          "upper": 17.379,
          "label": "±1 standard error"
        },
        "sourceUrl": "https://www.vals.ai/benchmarks/legal_research",
        "details": [
          {
            "label": "Publisher model id",
            "value": "alibaba/qwen3.6-plus"
          },
          {
            "label": "Standard error",
            "value": "±2.48 points"
          },
          {
            "label": "Mean latency",
            "value": "659 seconds per test"
          },
          {
            "label": "Evaluation mode",
            "value": "Agentic · many steps allowed"
          },
          {
            "label": "Administrative / Regulatory",
            "value": "19.7%"
          },
          {
            "label": "Business & Commercial",
            "value": "16.3%"
          },
          {
            "label": "Civil Litigation",
            "value": "14.8%"
          },
          {
            "label": "Constitutional / Civil Rights",
            "value": "10.5%"
          },
          {
            "label": "Criminal",
            "value": "10.5%"
          },
          {
            "label": "Family",
            "value": "0.0%"
          },
          {
            "label": "Health",
            "value": "30.0%"
          },
          {
            "label": "Immigration",
            "value": "18.2%"
          },
          {
            "label": "Weighted Pass",
            "value": "74.3%"
          }
        ]
      },
      {
        "id": "alibaba-qwen3-7-max",
        "label": "alibaba/qwen3.7-max",
        "model": "alibaba/qwen3.7-max",
        "provider": "Alibaba",
        "harness": null,
        "effort": null,
        "score": 25.481,
        "costUsd": 1.400701,
        "uncertainty": {
          "lower": 22.452,
          "upper": 28.51,
          "label": "±1 standard error"
        },
        "sourceUrl": "https://www.vals.ai/benchmarks/legal_research",
        "details": [
          {
            "label": "Publisher model id",
            "value": "alibaba/qwen3.7-max"
          },
          {
            "label": "Standard error",
            "value": "±3.03 points"
          },
          {
            "label": "Mean latency",
            "value": "809 seconds per test"
          },
          {
            "label": "Evaluation mode",
            "value": "Agentic · many steps allowed"
          },
          {
            "label": "Administrative / Regulatory",
            "value": "36.4%"
          },
          {
            "label": "Business & Commercial",
            "value": "20.9%"
          },
          {
            "label": "Civil Litigation",
            "value": "25.9%"
          },
          {
            "label": "Constitutional / Civil Rights",
            "value": "23.7%"
          },
          {
            "label": "Criminal",
            "value": "21.1%"
          },
          {
            "label": "Family",
            "value": "9.1%"
          },
          {
            "label": "Health",
            "value": "40.0%"
          },
          {
            "label": "Immigration",
            "value": "18.2%"
          },
          {
            "label": "Weighted Pass",
            "value": "78.4%"
          }
        ]
      },
      {
        "id": "alibaba-qwen3-7-plus",
        "label": "alibaba/qwen3.7-plus",
        "model": "alibaba/qwen3.7-plus",
        "provider": "Alibaba",
        "harness": null,
        "effort": null,
        "score": 16.346,
        "costUsd": 0.295949,
        "uncertainty": {
          "lower": 13.776,
          "upper": 18.916,
          "label": "±1 standard error"
        },
        "sourceUrl": "https://www.vals.ai/benchmarks/legal_research",
        "details": [
          {
            "label": "Publisher model id",
            "value": "alibaba/qwen3.7-plus"
          },
          {
            "label": "Standard error",
            "value": "±2.57 points"
          },
          {
            "label": "Mean latency",
            "value": "832 seconds per test"
          },
          {
            "label": "Evaluation mode",
            "value": "Agentic · many steps allowed"
          },
          {
            "label": "Administrative / Regulatory",
            "value": "15.2%"
          },
          {
            "label": "Business & Commercial",
            "value": "14.0%"
          },
          {
            "label": "Civil Litigation",
            "value": "25.9%"
          },
          {
            "label": "Constitutional / Civil Rights",
            "value": "21.1%"
          },
          {
            "label": "Criminal",
            "value": "15.8%"
          },
          {
            "label": "Family",
            "value": "0.0%"
          },
          {
            "label": "Health",
            "value": "26.7%"
          },
          {
            "label": "Immigration",
            "value": "18.2%"
          },
          {
            "label": "Weighted Pass",
            "value": "75.0%"
          }
        ]
      },
      {
        "id": "alibaba-qwen3-8-27b",
        "label": "alibaba/qwen3.8-27b",
        "model": "alibaba/qwen3.8-27b",
        "provider": "Alibaba",
        "harness": null,
        "effort": "xhigh",
        "score": 36.058,
        "costUsd": 2.159719,
        "uncertainty": {
          "lower": 32.721,
          "upper": 39.395,
          "label": "±1 standard error"
        },
        "sourceUrl": "https://www.vals.ai/benchmarks/legal_research",
        "details": [
          {
            "label": "Publisher model id",
            "value": "alibaba/qwen3.8-27b"
          },
          {
            "label": "Standard error",
            "value": "±3.34 points"
          },
          {
            "label": "Mean latency",
            "value": "1,601 seconds per test"
          },
          {
            "label": "Evaluation mode",
            "value": "Agentic · many steps allowed"
          },
          {
            "label": "Administrative / Regulatory",
            "value": "45.5%"
          },
          {
            "label": "Business & Commercial",
            "value": "41.9%"
          },
          {
            "label": "Civil Litigation",
            "value": "25.9%"
          },
          {
            "label": "Constitutional / Civil Rights",
            "value": "26.3%"
          },
          {
            "label": "Criminal",
            "value": "26.3%"
          },
          {
            "label": "Family",
            "value": "22.7%"
          },
          {
            "label": "Health",
            "value": "63.3%"
          },
          {
            "label": "Immigration",
            "value": "18.2%"
          },
          {
            "label": "Weighted Pass",
            "value": "82.6%"
          }
        ]
      },
      {
        "id": "alibaba-qwen3-8-max",
        "label": "alibaba/qwen3.8-max",
        "model": "alibaba/qwen3.8-max",
        "provider": "Alibaba",
        "harness": null,
        "effort": null,
        "score": 47.596,
        "costUsd": 2.486351,
        "uncertainty": {
          "lower": 44.125,
          "upper": 51.06699999999999,
          "label": "±1 standard error"
        },
        "sourceUrl": "https://www.vals.ai/benchmarks/legal_research",
        "details": [
          {
            "label": "Publisher model id",
            "value": "alibaba/qwen3.8-max"
          },
          {
            "label": "Standard error",
            "value": "±3.47 points"
          },
          {
            "label": "Mean latency",
            "value": "3,678 seconds per test"
          },
          {
            "label": "Evaluation mode",
            "value": "Agentic · many steps allowed"
          },
          {
            "label": "Administrative / Regulatory",
            "value": "59.1%"
          },
          {
            "label": "Business & Commercial",
            "value": "46.5%"
          },
          {
            "label": "Civil Litigation",
            "value": "48.1%"
          },
          {
            "label": "Constitutional / Civil Rights",
            "value": "34.2%"
          },
          {
            "label": "Criminal",
            "value": "40.4%"
          },
          {
            "label": "Family",
            "value": "18.2%"
          },
          {
            "label": "Health",
            "value": "73.3%"
          },
          {
            "label": "Immigration",
            "value": "45.5%"
          },
          {
            "label": "Weighted Pass",
            "value": "86.1%"
          }
        ]
      },
      {
        "id": "ant-ling-3-0-flash-2607",
        "label": "ant/ling-3.0-flash-2607",
        "model": "ant/ling-3.0-flash-2607",
        "provider": "Ant",
        "harness": null,
        "effort": null,
        "score": 0,
        "costUsd": 0.036023,
        "uncertainty": {
          "lower": 0,
          "upper": 0,
          "label": "±1 standard error"
        },
        "sourceUrl": "https://www.vals.ai/benchmarks/legal_research",
        "details": [
          {
            "label": "Publisher model id",
            "value": "ant/ling-3.0-flash-2607"
          },
          {
            "label": "Standard error",
            "value": "±0.00 points"
          },
          {
            "label": "Mean latency",
            "value": "212 seconds per test"
          },
          {
            "label": "Evaluation mode",
            "value": "Agentic · many steps allowed"
          },
          {
            "label": "Administrative / Regulatory",
            "value": "0.0%"
          },
          {
            "label": "Business & Commercial",
            "value": "0.0%"
          },
          {
            "label": "Civil Litigation",
            "value": "0.0%"
          },
          {
            "label": "Constitutional / Civil Rights",
            "value": "0.0%"
          },
          {
            "label": "Criminal",
            "value": "0.0%"
          },
          {
            "label": "Family",
            "value": "0.0%"
          },
          {
            "label": "Health",
            "value": "0.0%"
          },
          {
            "label": "Immigration",
            "value": "0.0%"
          },
          {
            "label": "Weighted Pass",
            "value": "64.6%"
          }
        ]
      },
      {
        "id": "ant-ling-3-0-flash-af-rc3",
        "label": "ant/ling-3.0-flash-af-rc3",
        "model": "ant/ling-3.0-flash-af-rc3",
        "provider": "Ant",
        "harness": null,
        "effort": null,
        "score": 15.385,
        "costUsd": 0.028227,
        "uncertainty": {
          "lower": 12.876999999999999,
          "upper": 17.893,
          "label": "±1 standard error"
        },
        "sourceUrl": "https://www.vals.ai/benchmarks/legal_research",
        "details": [
          {
            "label": "Publisher model id",
            "value": "ant/ling-3.0-flash-af-rc3"
          },
          {
            "label": "Standard error",
            "value": "±2.51 points"
          },
          {
            "label": "Mean latency",
            "value": "515 seconds per test"
          },
          {
            "label": "Evaluation mode",
            "value": "Agentic · many steps allowed"
          },
          {
            "label": "Administrative / Regulatory",
            "value": "18.2%"
          },
          {
            "label": "Business & Commercial",
            "value": "14.0%"
          },
          {
            "label": "Civil Litigation",
            "value": "22.2%"
          },
          {
            "label": "Constitutional / Civil Rights",
            "value": "10.5%"
          },
          {
            "label": "Criminal",
            "value": "12.3%"
          },
          {
            "label": "Family",
            "value": "4.5%"
          },
          {
            "label": "Health",
            "value": "26.7%"
          },
          {
            "label": "Immigration",
            "value": "27.3%"
          },
          {
            "label": "Weighted Pass",
            "value": "73.2%"
          }
        ]
      },
      {
        "id": "anthropic-claude-fable-5",
        "label": "anthropic/claude-fable-5",
        "model": "anthropic/claude-fable-5",
        "provider": "Anthropic",
        "harness": null,
        "effort": "max",
        "score": 49.519,
        "costUsd": 9.791426,
        "uncertainty": {
          "lower": 46.044,
          "upper": 52.994,
          "label": "±1 standard error"
        },
        "sourceUrl": "https://www.vals.ai/benchmarks/legal_research",
        "details": [
          {
            "label": "Publisher model id",
            "value": "anthropic/claude-fable-5"
          },
          {
            "label": "Standard error",
            "value": "±3.48 points"
          },
          {
            "label": "Mean latency",
            "value": "1,338 seconds per test"
          },
          {
            "label": "Evaluation mode",
            "value": "Agentic · many steps allowed"
          },
          {
            "label": "Administrative / Regulatory",
            "value": "57.6%"
          },
          {
            "label": "Business & Commercial",
            "value": "51.2%"
          },
          {
            "label": "Civil Litigation",
            "value": "44.4%"
          },
          {
            "label": "Constitutional / Civil Rights",
            "value": "50.0%"
          },
          {
            "label": "Criminal",
            "value": "42.1%"
          },
          {
            "label": "Family",
            "value": "36.4%"
          },
          {
            "label": "Health",
            "value": "66.7%"
          },
          {
            "label": "Immigration",
            "value": "36.4%"
          },
          {
            "label": "Weighted Pass",
            "value": "87.8%"
          }
        ]
      },
      {
        "id": "anthropic-claude-fable-5-1",
        "label": "anthropic/claude-fable-5-1",
        "model": "anthropic/claude-fable-5-1",
        "provider": "Anthropic",
        "harness": null,
        "effort": "max",
        "score": 55.288,
        "costUsd": 23.055789,
        "uncertainty": {
          "lower": 51.831999999999994,
          "upper": 58.744,
          "label": "±1 standard error"
        },
        "sourceUrl": "https://www.vals.ai/benchmarks/legal_research",
        "details": [
          {
            "label": "Publisher model id",
            "value": "anthropic/claude-fable-5-1"
          },
          {
            "label": "Standard error",
            "value": "±3.46 points"
          },
          {
            "label": "Mean latency",
            "value": "3,449 seconds per test"
          },
          {
            "label": "Evaluation mode",
            "value": "Agentic · many steps allowed"
          },
          {
            "label": "Administrative / Regulatory",
            "value": "68.2%"
          },
          {
            "label": "Business & Commercial",
            "value": "62.8%"
          },
          {
            "label": "Civil Litigation",
            "value": "63.0%"
          },
          {
            "label": "Constitutional / Civil Rights",
            "value": "42.1%"
          },
          {
            "label": "Criminal",
            "value": "38.6%"
          },
          {
            "label": "Family",
            "value": "31.8%"
          },
          {
            "label": "Health",
            "value": "80.0%"
          },
          {
            "label": "Immigration",
            "value": "36.4%"
          },
          {
            "label": "Weighted Pass",
            "value": "90.2%"
          }
        ]
      },
      {
        "id": "anthropic-claude-haiku-4-5-20251001-thinking",
        "label": "anthropic/claude-haiku-4-5-20251001-thinking",
        "model": "anthropic/claude-haiku-4-5-20251001-thinking",
        "provider": "Anthropic",
        "harness": null,
        "effort": null,
        "score": 10.577,
        "costUsd": 0.26891,
        "uncertainty": {
          "lower": 8.439,
          "upper": 12.715,
          "label": "±1 standard error"
        },
        "sourceUrl": "https://www.vals.ai/benchmarks/legal_research",
        "details": [
          {
            "label": "Publisher model id",
            "value": "anthropic/claude-haiku-4-5-20251001-thinking"
          },
          {
            "label": "Standard error",
            "value": "±2.14 points"
          },
          {
            "label": "Mean latency",
            "value": "243 seconds per test"
          },
          {
            "label": "Evaluation mode",
            "value": "Agentic · many steps allowed"
          },
          {
            "label": "Administrative / Regulatory",
            "value": "10.6%"
          },
          {
            "label": "Business & Commercial",
            "value": "9.3%"
          },
          {
            "label": "Civil Litigation",
            "value": "7.4%"
          },
          {
            "label": "Constitutional / Civil Rights",
            "value": "13.2%"
          },
          {
            "label": "Criminal",
            "value": "10.5%"
          },
          {
            "label": "Family",
            "value": "0.0%"
          },
          {
            "label": "Health",
            "value": "6.7%"
          },
          {
            "label": "Immigration",
            "value": "27.3%"
          },
          {
            "label": "Weighted Pass",
            "value": "64.9%"
          }
        ]
      },
      {
        "id": "anthropic-claude-opus-4-7",
        "label": "anthropic/claude-opus-4-7",
        "model": "anthropic/claude-opus-4-7",
        "provider": "Anthropic",
        "harness": null,
        "effort": "max",
        "score": 38.462,
        "costUsd": 3.05609,
        "uncertainty": {
          "lower": 35.081,
          "upper": 41.843,
          "label": "±1 standard error"
        },
        "sourceUrl": "https://www.vals.ai/benchmarks/legal_research",
        "details": [
          {
            "label": "Publisher model id",
            "value": "anthropic/claude-opus-4-7"
          },
          {
            "label": "Standard error",
            "value": "±3.38 points"
          },
          {
            "label": "Mean latency",
            "value": "512 seconds per test"
          },
          {
            "label": "Evaluation mode",
            "value": "Agentic · many steps allowed"
          },
          {
            "label": "Administrative / Regulatory",
            "value": "45.5%"
          },
          {
            "label": "Business & Commercial",
            "value": "32.6%"
          },
          {
            "label": "Civil Litigation",
            "value": "37.0%"
          },
          {
            "label": "Constitutional / Civil Rights",
            "value": "44.7%"
          },
          {
            "label": "Criminal",
            "value": "29.8%"
          },
          {
            "label": "Family",
            "value": "31.8%"
          },
          {
            "label": "Health",
            "value": "53.3%"
          },
          {
            "label": "Immigration",
            "value": "36.4%"
          },
          {
            "label": "Weighted Pass",
            "value": "84.1%"
          }
        ]
      },
      {
        "id": "anthropic-claude-opus-4-8",
        "label": "anthropic/claude-opus-4-8",
        "model": "anthropic/claude-opus-4-8",
        "provider": "Anthropic",
        "harness": null,
        "effort": "max",
        "score": 43.75,
        "costUsd": 2.820942,
        "uncertainty": {
          "lower": 40.302,
          "upper": 47.198,
          "label": "±1 standard error"
        },
        "sourceUrl": "https://www.vals.ai/benchmarks/legal_research",
        "details": [
          {
            "label": "Publisher model id",
            "value": "anthropic/claude-opus-4-8"
          },
          {
            "label": "Standard error",
            "value": "±3.45 points"
          },
          {
            "label": "Mean latency",
            "value": "714 seconds per test"
          },
          {
            "label": "Evaluation mode",
            "value": "Agentic · many steps allowed"
          },
          {
            "label": "Administrative / Regulatory",
            "value": "43.9%"
          },
          {
            "label": "Business & Commercial",
            "value": "41.9%"
          },
          {
            "label": "Civil Litigation",
            "value": "51.9%"
          },
          {
            "label": "Constitutional / Civil Rights",
            "value": "39.5%"
          },
          {
            "label": "Criminal",
            "value": "45.6%"
          },
          {
            "label": "Family",
            "value": "18.2%"
          },
          {
            "label": "Health",
            "value": "53.3%"
          },
          {
            "label": "Immigration",
            "value": "27.3%"
          },
          {
            "label": "Weighted Pass",
            "value": "86.1%"
          }
        ]
      },
      {
        "id": "anthropic-claude-opus-5",
        "label": "anthropic/claude-opus-5",
        "model": "anthropic/claude-opus-5",
        "provider": "Anthropic",
        "harness": null,
        "effort": "max",
        "score": 55.288,
        "costUsd": 6.762498,
        "uncertainty": {
          "lower": 51.831999999999994,
          "upper": 58.744,
          "label": "±1 standard error"
        },
        "sourceUrl": "https://www.vals.ai/benchmarks/legal_research",
        "details": [
          {
            "label": "Publisher model id",
            "value": "anthropic/claude-opus-5"
          },
          {
            "label": "Standard error",
            "value": "±3.46 points"
          },
          {
            "label": "Mean latency",
            "value": "1,861 seconds per test"
          },
          {
            "label": "Evaluation mode",
            "value": "Agentic · many steps allowed"
          },
          {
            "label": "Administrative / Regulatory",
            "value": "65.2%"
          },
          {
            "label": "Business & Commercial",
            "value": "62.8%"
          },
          {
            "label": "Civil Litigation",
            "value": "63.0%"
          },
          {
            "label": "Constitutional / Civil Rights",
            "value": "44.7%"
          },
          {
            "label": "Criminal",
            "value": "43.9%"
          },
          {
            "label": "Family",
            "value": "40.9%"
          },
          {
            "label": "Health",
            "value": "76.7%"
          },
          {
            "label": "Immigration",
            "value": "27.3%"
          },
          {
            "label": "Weighted Pass",
            "value": "90.6%"
          }
        ]
      },
      {
        "id": "anthropic-claude-opus-5-5",
        "label": "anthropic/claude-opus-5-5",
        "model": "anthropic/claude-opus-5-5",
        "provider": "Anthropic",
        "harness": null,
        "effort": "max",
        "score": 50.481,
        "costUsd": 25.163725,
        "uncertainty": {
          "lower": 47.006,
          "upper": 53.956,
          "label": "±1 standard error"
        },
        "sourceUrl": "https://www.vals.ai/benchmarks/legal_research",
        "details": [
          {
            "label": "Publisher model id",
            "value": "anthropic/claude-opus-5-5"
          },
          {
            "label": "Standard error",
            "value": "±3.48 points"
          },
          {
            "label": "Mean latency",
            "value": "6,750 seconds per test"
          },
          {
            "label": "Evaluation mode",
            "value": "Agentic · many steps allowed"
          },
          {
            "label": "Administrative / Regulatory",
            "value": "65.2%"
          },
          {
            "label": "Business & Commercial",
            "value": "48.8%"
          },
          {
            "label": "Civil Litigation",
            "value": "48.1%"
          },
          {
            "label": "Constitutional / Civil Rights",
            "value": "42.1%"
          },
          {
            "label": "Criminal",
            "value": "42.1%"
          },
          {
            "label": "Family",
            "value": "22.7%"
          },
          {
            "label": "Health",
            "value": "80.0%"
          },
          {
            "label": "Immigration",
            "value": "54.5%"
          },
          {
            "label": "Weighted Pass",
            "value": "80.4%"
          }
        ]
      },
      {
        "id": "anthropic-claude-sonnet-4-6",
        "label": "anthropic/claude-sonnet-4-6",
        "model": "anthropic/claude-sonnet-4-6",
        "provider": "Anthropic",
        "harness": null,
        "effort": "max",
        "score": 38.462,
        "costUsd": 2.302948,
        "uncertainty": {
          "lower": 35.081,
          "upper": 41.843,
          "label": "±1 standard error"
        },
        "sourceUrl": "https://www.vals.ai/benchmarks/legal_research",
        "details": [
          {
            "label": "Publisher model id",
            "value": "anthropic/claude-sonnet-4-6"
          },
          {
            "label": "Standard error",
            "value": "±3.38 points"
          },
          {
            "label": "Mean latency",
            "value": "990 seconds per test"
          },
          {
            "label": "Evaluation mode",
            "value": "Agentic · many steps allowed"
          },
          {
            "label": "Administrative / Regulatory",
            "value": "43.9%"
          },
          {
            "label": "Business & Commercial",
            "value": "34.9%"
          },
          {
            "label": "Civil Litigation",
            "value": "29.6%"
          },
          {
            "label": "Constitutional / Civil Rights",
            "value": "34.2%"
          },
          {
            "label": "Criminal",
            "value": "36.8%"
          },
          {
            "label": "Family",
            "value": "27.3%"
          },
          {
            "label": "Health",
            "value": "53.3%"
          },
          {
            "label": "Immigration",
            "value": "27.3%"
          },
          {
            "label": "Weighted Pass",
            "value": "85.3%"
          }
        ]
      },
      {
        "id": "anthropic-claude-sonnet-5",
        "label": "anthropic/claude-sonnet-5",
        "model": "anthropic/claude-sonnet-5",
        "provider": "Anthropic",
        "harness": null,
        "effort": "max",
        "score": 41.827,
        "costUsd": 2.719446,
        "uncertainty": {
          "lower": 38.399,
          "upper": 45.254999999999995,
          "label": "±1 standard error"
        },
        "sourceUrl": "https://www.vals.ai/benchmarks/legal_research",
        "details": [
          {
            "label": "Publisher model id",
            "value": "anthropic/claude-sonnet-5"
          },
          {
            "label": "Standard error",
            "value": "±3.43 points"
          },
          {
            "label": "Mean latency",
            "value": "1,546 seconds per test"
          },
          {
            "label": "Evaluation mode",
            "value": "Agentic · many steps allowed"
          },
          {
            "label": "Administrative / Regulatory",
            "value": "54.5%"
          },
          {
            "label": "Business & Commercial",
            "value": "32.6%"
          },
          {
            "label": "Civil Litigation",
            "value": "44.4%"
          },
          {
            "label": "Constitutional / Civil Rights",
            "value": "44.7%"
          },
          {
            "label": "Criminal",
            "value": "40.4%"
          },
          {
            "label": "Family",
            "value": "18.2%"
          },
          {
            "label": "Health",
            "value": "63.3%"
          },
          {
            "label": "Immigration",
            "value": "36.4%"
          },
          {
            "label": "Weighted Pass",
            "value": "84.6%"
          }
        ]
      },
      {
        "id": "anthropic-claude-sonnet-5-5",
        "label": "anthropic/claude-sonnet-5-5",
        "model": "anthropic/claude-sonnet-5-5",
        "provider": "Anthropic",
        "harness": null,
        "effort": "max",
        "score": 48.077,
        "costUsd": 14.826571,
        "uncertainty": {
          "lower": 44.604,
          "upper": 51.55,
          "label": "±1 standard error"
        },
        "sourceUrl": "https://www.vals.ai/benchmarks/legal_research",
        "details": [
          {
            "label": "Publisher model id",
            "value": "anthropic/claude-sonnet-5-5"
          },
          {
            "label": "Standard error",
            "value": "±3.47 points"
          },
          {
            "label": "Mean latency",
            "value": "5,589 seconds per test"
          },
          {
            "label": "Evaluation mode",
            "value": "Agentic · many steps allowed"
          },
          {
            "label": "Administrative / Regulatory",
            "value": "69.7%"
          },
          {
            "label": "Business & Commercial",
            "value": "39.5%"
          },
          {
            "label": "Civil Litigation",
            "value": "48.1%"
          },
          {
            "label": "Constitutional / Civil Rights",
            "value": "42.1%"
          },
          {
            "label": "Criminal",
            "value": "38.6%"
          },
          {
            "label": "Family",
            "value": "18.2%"
          },
          {
            "label": "Health",
            "value": "80.0%"
          },
          {
            "label": "Immigration",
            "value": "45.5%"
          },
          {
            "label": "Weighted Pass",
            "value": "84.3%"
          }
        ]
      },
      {
        "id": "deepseek-deepseek-v4-1-flash",
        "label": "deepseek/deepseek-v4.1-flash",
        "model": "deepseek/deepseek-v4.1-flash",
        "provider": "DeepSeek",
        "harness": null,
        "effort": "high",
        "score": 41.346,
        "costUsd": 0.24793,
        "uncertainty": {
          "lower": 37.922999999999995,
          "upper": 44.769,
          "label": "±1 standard error"
        },
        "sourceUrl": "https://www.vals.ai/benchmarks/legal_research",
        "details": [
          {
            "label": "Publisher model id",
            "value": "deepseek/deepseek-v4.1-flash"
          },
          {
            "label": "Standard error",
            "value": "±3.42 points"
          },
          {
            "label": "Mean latency",
            "value": "702 seconds per test"
          },
          {
            "label": "Evaluation mode",
            "value": "Agentic · many steps allowed"
          },
          {
            "label": "Administrative / Regulatory",
            "value": "53.0%"
          },
          {
            "label": "Business & Commercial",
            "value": "39.5%"
          },
          {
            "label": "Civil Litigation",
            "value": "48.1%"
          },
          {
            "label": "Constitutional / Civil Rights",
            "value": "36.8%"
          },
          {
            "label": "Criminal",
            "value": "33.3%"
          },
          {
            "label": "Family",
            "value": "13.6%"
          },
          {
            "label": "Health",
            "value": "73.3%"
          },
          {
            "label": "Immigration",
            "value": "36.4%"
          },
          {
            "label": "Weighted Pass",
            "value": "84.5%"
          }
        ]
      },
      {
        "id": "deepseek-deepseek-v4-flash-0731",
        "label": "deepseek/deepseek-v4-flash-0731",
        "model": "deepseek/deepseek-v4-flash-0731",
        "provider": "DeepSeek",
        "harness": null,
        "effort": "high",
        "score": 30.288,
        "costUsd": 0.275172,
        "uncertainty": {
          "lower": 27.094,
          "upper": 33.482,
          "label": "±1 standard error"
        },
        "sourceUrl": "https://www.vals.ai/benchmarks/legal_research",
        "details": [
          {
            "label": "Publisher model id",
            "value": "deepseek/deepseek-v4-flash-0731"
          },
          {
            "label": "Standard error",
            "value": "±3.19 points"
          },
          {
            "label": "Mean latency",
            "value": "1,267 seconds per test"
          },
          {
            "label": "Evaluation mode",
            "value": "Agentic · many steps allowed"
          },
          {
            "label": "Administrative / Regulatory",
            "value": "36.4%"
          },
          {
            "label": "Business & Commercial",
            "value": "23.3%"
          },
          {
            "label": "Civil Litigation",
            "value": "44.4%"
          },
          {
            "label": "Constitutional / Civil Rights",
            "value": "31.6%"
          },
          {
            "label": "Criminal",
            "value": "24.6%"
          },
          {
            "label": "Family",
            "value": "18.2%"
          },
          {
            "label": "Health",
            "value": "36.7%"
          },
          {
            "label": "Immigration",
            "value": "27.3%"
          },
          {
            "label": "Weighted Pass",
            "value": "81.2%"
          }
        ]
      },
      {
        "id": "deepseek-deepseek-v4-pro",
        "label": "deepseek/deepseek-v4-pro",
        "model": "deepseek/deepseek-v4-pro",
        "provider": "DeepSeek",
        "harness": null,
        "effort": "max",
        "score": 23.077,
        "costUsd": 0.682218,
        "uncertainty": {
          "lower": 20.149,
          "upper": 26.005000000000003,
          "label": "±1 standard error"
        },
        "sourceUrl": "https://www.vals.ai/benchmarks/legal_research",
        "details": [
          {
            "label": "Publisher model id",
            "value": "deepseek/deepseek-v4-pro"
          },
          {
            "label": "Standard error",
            "value": "±2.93 points"
          },
          {
            "label": "Mean latency",
            "value": "1,025 seconds per test"
          },
          {
            "label": "Evaluation mode",
            "value": "Agentic · many steps allowed"
          },
          {
            "label": "Administrative / Regulatory",
            "value": "28.8%"
          },
          {
            "label": "Business & Commercial",
            "value": "23.3%"
          },
          {
            "label": "Civil Litigation",
            "value": "33.3%"
          },
          {
            "label": "Constitutional / Civil Rights",
            "value": "13.2%"
          },
          {
            "label": "Criminal",
            "value": "15.8%"
          },
          {
            "label": "Family",
            "value": "9.1%"
          },
          {
            "label": "Health",
            "value": "46.7%"
          },
          {
            "label": "Immigration",
            "value": "9.1%"
          },
          {
            "label": "Weighted Pass",
            "value": "80.7%"
          }
        ]
      },
      {
        "id": "deepseek-deepseek-v4-pro-0813",
        "label": "deepseek/deepseek-v4-pro-0813",
        "model": "deepseek/deepseek-v4-pro-0813",
        "provider": "DeepSeek",
        "harness": null,
        "effort": "max",
        "score": 40.865,
        "costUsd": 1.088017,
        "uncertainty": {
          "lower": 37.448,
          "upper": 44.282000000000004,
          "label": "±1 standard error"
        },
        "sourceUrl": "https://www.vals.ai/benchmarks/legal_research",
        "details": [
          {
            "label": "Publisher model id",
            "value": "deepseek/deepseek-v4-pro-0813"
          },
          {
            "label": "Standard error",
            "value": "±3.42 points"
          },
          {
            "label": "Mean latency",
            "value": "2,527 seconds per test"
          },
          {
            "label": "Evaluation mode",
            "value": "Agentic · many steps allowed"
          },
          {
            "label": "Administrative / Regulatory",
            "value": "50.0%"
          },
          {
            "label": "Business & Commercial",
            "value": "34.9%"
          },
          {
            "label": "Civil Litigation",
            "value": "40.7%"
          },
          {
            "label": "Constitutional / Civil Rights",
            "value": "39.5%"
          },
          {
            "label": "Criminal",
            "value": "35.1%"
          },
          {
            "label": "Family",
            "value": "22.7%"
          },
          {
            "label": "Health",
            "value": "60.0%"
          },
          {
            "label": "Immigration",
            "value": "45.5%"
          },
          {
            "label": "Weighted Pass",
            "value": "83.8%"
          }
        ]
      },
      {
        "id": "fireworks-nemotron-lightning-3p5-30b-a3b",
        "label": "fireworks/nemotron-lightning-3p5-30b-a3b",
        "model": "fireworks/nemotron-lightning-3p5-30b-a3b",
        "provider": "Fireworks AI",
        "harness": null,
        "effort": null,
        "score": 2.885,
        "costUsd": 0.004365,
        "uncertainty": {
          "lower": 1.7219999999999998,
          "upper": 4.048,
          "label": "±1 standard error"
        },
        "sourceUrl": "https://www.vals.ai/benchmarks/legal_research",
        "details": [
          {
            "label": "Publisher model id",
            "value": "fireworks/nemotron-lightning-3p5-30b-a3b"
          },
          {
            "label": "Standard error",
            "value": "±1.16 points"
          },
          {
            "label": "Mean latency",
            "value": "285 seconds per test"
          },
          {
            "label": "Evaluation mode",
            "value": "Agentic · many steps allowed"
          },
          {
            "label": "Administrative / Regulatory",
            "value": "1.5%"
          },
          {
            "label": "Business & Commercial",
            "value": "4.7%"
          },
          {
            "label": "Civil Litigation",
            "value": "3.7%"
          },
          {
            "label": "Constitutional / Civil Rights",
            "value": "5.3%"
          },
          {
            "label": "Criminal",
            "value": "3.5%"
          },
          {
            "label": "Family",
            "value": "0.0%"
          },
          {
            "label": "Health",
            "value": "0.0%"
          },
          {
            "label": "Immigration",
            "value": "0.0%"
          },
          {
            "label": "Weighted Pass",
            "value": "49.0%"
          }
        ]
      },
      {
        "id": "google-gemini-3-1-flash-lite-preview",
        "label": "google/gemini-3.1-flash-lite-preview",
        "model": "google/gemini-3.1-flash-lite-preview",
        "provider": "Google",
        "harness": null,
        "effort": null,
        "score": 3.365,
        "costUsd": 0.013607,
        "uncertainty": {
          "lower": 2.112,
          "upper": 4.618,
          "label": "±1 standard error"
        },
        "sourceUrl": "https://www.vals.ai/benchmarks/legal_research",
        "details": [
          {
            "label": "Publisher model id",
            "value": "google/gemini-3.1-flash-lite-preview"
          },
          {
            "label": "Standard error",
            "value": "±1.25 points"
          },
          {
            "label": "Mean latency",
            "value": "20 seconds per test"
          },
          {
            "label": "Evaluation mode",
            "value": "Agentic · many steps allowed"
          },
          {
            "label": "Administrative / Regulatory",
            "value": "3.0%"
          },
          {
            "label": "Business & Commercial",
            "value": "2.3%"
          },
          {
            "label": "Civil Litigation",
            "value": "3.7%"
          },
          {
            "label": "Constitutional / Civil Rights",
            "value": "2.6%"
          },
          {
            "label": "Criminal",
            "value": "1.8%"
          },
          {
            "label": "Family",
            "value": "0.0%"
          },
          {
            "label": "Health",
            "value": "6.7%"
          },
          {
            "label": "Immigration",
            "value": "9.1%"
          },
          {
            "label": "Weighted Pass",
            "value": "54.4%"
          }
        ]
      },
      {
        "id": "google-gemini-3-1-pro-preview",
        "label": "google/gemini-3.1-pro-preview",
        "model": "google/gemini-3.1-pro-preview",
        "provider": "Google",
        "harness": null,
        "effort": "high",
        "score": 20.673,
        "costUsd": 1.100145,
        "uncertainty": {
          "lower": 17.857999999999997,
          "upper": 23.488,
          "label": "±1 standard error"
        },
        "sourceUrl": "https://www.vals.ai/benchmarks/legal_research",
        "details": [
          {
            "label": "Publisher model id",
            "value": "google/gemini-3.1-pro-preview"
          },
          {
            "label": "Standard error",
            "value": "±2.81 points"
          },
          {
            "label": "Mean latency",
            "value": "242 seconds per test"
          },
          {
            "label": "Evaluation mode",
            "value": "Agentic · many steps allowed"
          },
          {
            "label": "Administrative / Regulatory",
            "value": "24.2%"
          },
          {
            "label": "Business & Commercial",
            "value": "23.3%"
          },
          {
            "label": "Civil Litigation",
            "value": "11.1%"
          },
          {
            "label": "Constitutional / Civil Rights",
            "value": "21.1%"
          },
          {
            "label": "Criminal",
            "value": "19.3%"
          },
          {
            "label": "Family",
            "value": "4.5%"
          },
          {
            "label": "Health",
            "value": "33.3%"
          },
          {
            "label": "Immigration",
            "value": "18.2%"
          },
          {
            "label": "Weighted Pass",
            "value": "76.6%"
          }
        ]
      },
      {
        "id": "google-gemini-3-5-flash",
        "label": "google/gemini-3.5-flash",
        "model": "google/gemini-3.5-flash",
        "provider": "Google",
        "harness": null,
        "effort": "high",
        "score": 30.769,
        "costUsd": 1.305281,
        "uncertainty": {
          "lower": 27.561,
          "upper": 33.977,
          "label": "±1 standard error"
        },
        "sourceUrl": "https://www.vals.ai/benchmarks/legal_research",
        "details": [
          {
            "label": "Publisher model id",
            "value": "google/gemini-3.5-flash"
          },
          {
            "label": "Standard error",
            "value": "±3.21 points"
          },
          {
            "label": "Mean latency",
            "value": "278 seconds per test"
          },
          {
            "label": "Evaluation mode",
            "value": "Agentic · many steps allowed"
          },
          {
            "label": "Administrative / Regulatory",
            "value": "43.9%"
          },
          {
            "label": "Business & Commercial",
            "value": "27.9%"
          },
          {
            "label": "Civil Litigation",
            "value": "22.2%"
          },
          {
            "label": "Constitutional / Civil Rights",
            "value": "28.9%"
          },
          {
            "label": "Criminal",
            "value": "26.3%"
          },
          {
            "label": "Family",
            "value": "18.2%"
          },
          {
            "label": "Health",
            "value": "43.3%"
          },
          {
            "label": "Immigration",
            "value": "18.2%"
          },
          {
            "label": "Weighted Pass",
            "value": "83.9%"
          }
        ]
      },
      {
        "id": "google-gemini-3-5-flash-lite",
        "label": "google/gemini-3.5-flash-lite",
        "model": "google/gemini-3.5-flash-lite",
        "provider": "Google",
        "harness": null,
        "effort": "high",
        "score": 13.942,
        "costUsd": 0.124723,
        "uncertainty": {
          "lower": 11.534,
          "upper": 16.35,
          "label": "±1 standard error"
        },
        "sourceUrl": "https://www.vals.ai/benchmarks/legal_research",
        "details": [
          {
            "label": "Publisher model id",
            "value": "google/gemini-3.5-flash-lite"
          },
          {
            "label": "Standard error",
            "value": "±2.41 points"
          },
          {
            "label": "Mean latency",
            "value": "59 seconds per test"
          },
          {
            "label": "Evaluation mode",
            "value": "Agentic · many steps allowed"
          },
          {
            "label": "Administrative / Regulatory",
            "value": "19.7%"
          },
          {
            "label": "Business & Commercial",
            "value": "7.0%"
          },
          {
            "label": "Civil Litigation",
            "value": "11.1%"
          },
          {
            "label": "Constitutional / Civil Rights",
            "value": "21.1%"
          },
          {
            "label": "Criminal",
            "value": "12.3%"
          },
          {
            "label": "Family",
            "value": "4.5%"
          },
          {
            "label": "Health",
            "value": "13.3%"
          },
          {
            "label": "Immigration",
            "value": "18.2%"
          },
          {
            "label": "Weighted Pass",
            "value": "68.9%"
          }
        ]
      },
      {
        "id": "google-gemini-3-6-flash",
        "label": "google/gemini-3.6-flash",
        "model": "google/gemini-3.6-flash",
        "provider": "Google",
        "harness": null,
        "effort": "high",
        "score": 25,
        "costUsd": 0.466741,
        "uncertainty": {
          "lower": 21.990000000000002,
          "upper": 28.009999999999998,
          "label": "±1 standard error"
        },
        "sourceUrl": "https://www.vals.ai/benchmarks/legal_research",
        "details": [
          {
            "label": "Publisher model id",
            "value": "google/gemini-3.6-flash"
          },
          {
            "label": "Standard error",
            "value": "±3.01 points"
          },
          {
            "label": "Mean latency",
            "value": "168 seconds per test"
          },
          {
            "label": "Evaluation mode",
            "value": "Agentic · many steps allowed"
          },
          {
            "label": "Administrative / Regulatory",
            "value": "31.8%"
          },
          {
            "label": "Business & Commercial",
            "value": "23.3%"
          },
          {
            "label": "Civil Litigation",
            "value": "18.5%"
          },
          {
            "label": "Constitutional / Civil Rights",
            "value": "18.4%"
          },
          {
            "label": "Criminal",
            "value": "24.6%"
          },
          {
            "label": "Family",
            "value": "9.1%"
          },
          {
            "label": "Health",
            "value": "36.7%"
          },
          {
            "label": "Immigration",
            "value": "18.2%"
          },
          {
            "label": "Weighted Pass",
            "value": "80.5%"
          }
        ]
      },
      {
        "id": "google-gemini-3-7-flash",
        "label": "google/gemini-3.7-flash",
        "model": "google/gemini-3.7-flash",
        "provider": "Google",
        "harness": null,
        "effort": "high",
        "score": 34.615,
        "costUsd": 0.760397,
        "uncertainty": {
          "lower": 31.308000000000003,
          "upper": 37.922000000000004,
          "label": "±1 standard error"
        },
        "sourceUrl": "https://www.vals.ai/benchmarks/legal_research",
        "details": [
          {
            "label": "Publisher model id",
            "value": "google/gemini-3.7-flash"
          },
          {
            "label": "Standard error",
            "value": "±3.31 points"
          },
          {
            "label": "Mean latency",
            "value": "289 seconds per test"
          },
          {
            "label": "Evaluation mode",
            "value": "Agentic · many steps allowed"
          },
          {
            "label": "Administrative / Regulatory",
            "value": "42.4%"
          },
          {
            "label": "Business & Commercial",
            "value": "37.2%"
          },
          {
            "label": "Civil Litigation",
            "value": "29.6%"
          },
          {
            "label": "Constitutional / Civil Rights",
            "value": "31.6%"
          },
          {
            "label": "Criminal",
            "value": "33.3%"
          },
          {
            "label": "Family",
            "value": "13.6%"
          },
          {
            "label": "Health",
            "value": "50.0%"
          },
          {
            "label": "Immigration",
            "value": "36.4%"
          },
          {
            "label": "Weighted Pass",
            "value": "84.1%"
          }
        ]
      },
      {
        "id": "google-gemini-3-8-flash",
        "label": "google/gemini-3.8-flash",
        "model": "google/gemini-3.8-flash",
        "provider": "Google",
        "harness": null,
        "effort": "high",
        "score": 38.942,
        "costUsd": 1.630498,
        "uncertainty": {
          "lower": 35.553,
          "upper": 42.331,
          "label": "±1 standard error"
        },
        "sourceUrl": "https://www.vals.ai/benchmarks/legal_research",
        "details": [
          {
            "label": "Publisher model id",
            "value": "google/gemini-3.8-flash"
          },
          {
            "label": "Standard error",
            "value": "±3.39 points"
          },
          {
            "label": "Mean latency",
            "value": "324 seconds per test"
          },
          {
            "label": "Evaluation mode",
            "value": "Agentic · many steps allowed"
          },
          {
            "label": "Administrative / Regulatory",
            "value": "47.0%"
          },
          {
            "label": "Business & Commercial",
            "value": "41.9%"
          },
          {
            "label": "Civil Litigation",
            "value": "29.6%"
          },
          {
            "label": "Constitutional / Civil Rights",
            "value": "42.1%"
          },
          {
            "label": "Criminal",
            "value": "40.4%"
          },
          {
            "label": "Family",
            "value": "18.2%"
          },
          {
            "label": "Health",
            "value": "56.7%"
          },
          {
            "label": "Immigration",
            "value": "27.3%"
          },
          {
            "label": "Weighted Pass",
            "value": "87.5%"
          }
        ]
      },
      {
        "id": "google-gemini-3-flash-preview",
        "label": "google/gemini-3-flash-preview",
        "model": "google/gemini-3-flash-preview",
        "provider": "Google",
        "harness": null,
        "effort": "high",
        "score": 18.269,
        "costUsd": 0.135395,
        "uncertainty": {
          "lower": 15.582999999999998,
          "upper": 20.955,
          "label": "±1 standard error"
        },
        "sourceUrl": "https://www.vals.ai/benchmarks/legal_research",
        "details": [
          {
            "label": "Publisher model id",
            "value": "google/gemini-3-flash-preview"
          },
          {
            "label": "Standard error",
            "value": "±2.69 points"
          },
          {
            "label": "Mean latency",
            "value": "188 seconds per test"
          },
          {
            "label": "Evaluation mode",
            "value": "Agentic · many steps allowed"
          },
          {
            "label": "Administrative / Regulatory",
            "value": "27.3%"
          },
          {
            "label": "Business & Commercial",
            "value": "18.6%"
          },
          {
            "label": "Civil Litigation",
            "value": "11.1%"
          },
          {
            "label": "Constitutional / Civil Rights",
            "value": "7.9%"
          },
          {
            "label": "Criminal",
            "value": "12.3%"
          },
          {
            "label": "Family",
            "value": "4.5%"
          },
          {
            "label": "Health",
            "value": "30.0%"
          },
          {
            "label": "Immigration",
            "value": "27.3%"
          },
          {
            "label": "Weighted Pass",
            "value": "75.2%"
          }
        ]
      },
      {
        "id": "grok-grok-4-20-0309-reasoning",
        "label": "grok/grok-4.20-0309-reasoning",
        "model": "grok/grok-4.20-0309-reasoning",
        "provider": "SpaceXAI",
        "harness": null,
        "effort": null,
        "score": 13.942,
        "costUsd": 0.261499,
        "uncertainty": {
          "lower": 11.534,
          "upper": 16.35,
          "label": "±1 standard error"
        },
        "sourceUrl": "https://www.vals.ai/benchmarks/legal_research",
        "details": [
          {
            "label": "Publisher model id",
            "value": "grok/grok-4.20-0309-reasoning"
          },
          {
            "label": "Standard error",
            "value": "±2.41 points"
          },
          {
            "label": "Mean latency",
            "value": "244 seconds per test"
          },
          {
            "label": "Evaluation mode",
            "value": "Agentic · many steps allowed"
          },
          {
            "label": "Administrative / Regulatory",
            "value": "9.1%"
          },
          {
            "label": "Business & Commercial",
            "value": "16.3%"
          },
          {
            "label": "Civil Litigation",
            "value": "25.9%"
          },
          {
            "label": "Constitutional / Civil Rights",
            "value": "18.4%"
          },
          {
            "label": "Criminal",
            "value": "17.5%"
          },
          {
            "label": "Family",
            "value": "0.0%"
          },
          {
            "label": "Health",
            "value": "10.0%"
          },
          {
            "label": "Immigration",
            "value": "18.2%"
          },
          {
            "label": "Weighted Pass",
            "value": "65.1%"
          }
        ]
      },
      {
        "id": "grok-grok-4-3",
        "label": "grok/grok-4.3",
        "model": "grok/grok-4.3",
        "provider": "SpaceXAI",
        "harness": null,
        "effort": "high",
        "score": 15.385,
        "costUsd": 0.203143,
        "uncertainty": {
          "lower": 12.876999999999999,
          "upper": 17.893,
          "label": "±1 standard error"
        },
        "sourceUrl": "https://www.vals.ai/benchmarks/legal_research",
        "details": [
          {
            "label": "Publisher model id",
            "value": "grok/grok-4.3"
          },
          {
            "label": "Standard error",
            "value": "±2.51 points"
          },
          {
            "label": "Mean latency",
            "value": "85 seconds per test"
          },
          {
            "label": "Evaluation mode",
            "value": "Agentic · many steps allowed"
          },
          {
            "label": "Administrative / Regulatory",
            "value": "25.8%"
          },
          {
            "label": "Business & Commercial",
            "value": "14.0%"
          },
          {
            "label": "Civil Litigation",
            "value": "3.7%"
          },
          {
            "label": "Constitutional / Civil Rights",
            "value": "13.2%"
          },
          {
            "label": "Criminal",
            "value": "14.0%"
          },
          {
            "label": "Family",
            "value": "0.0%"
          },
          {
            "label": "Health",
            "value": "40.0%"
          },
          {
            "label": "Immigration",
            "value": "9.1%"
          },
          {
            "label": "Weighted Pass",
            "value": "69.5%"
          }
        ]
      },
      {
        "id": "grok-grok-4-5",
        "label": "grok/grok-4.5",
        "model": "grok/grok-4.5",
        "provider": "SpaceXAI",
        "harness": null,
        "effort": "high",
        "score": 37.981,
        "costUsd": 1.086814,
        "uncertainty": {
          "lower": 34.608000000000004,
          "upper": 41.354,
          "label": "±1 standard error"
        },
        "sourceUrl": "https://www.vals.ai/benchmarks/legal_research",
        "details": [
          {
            "label": "Publisher model id",
            "value": "grok/grok-4.5"
          },
          {
            "label": "Standard error",
            "value": "±3.37 points"
          },
          {
            "label": "Mean latency",
            "value": "954 seconds per test"
          },
          {
            "label": "Evaluation mode",
            "value": "Agentic · many steps allowed"
          },
          {
            "label": "Administrative / Regulatory",
            "value": "51.5%"
          },
          {
            "label": "Business & Commercial",
            "value": "30.2%"
          },
          {
            "label": "Civil Litigation",
            "value": "33.3%"
          },
          {
            "label": "Constitutional / Civil Rights",
            "value": "31.6%"
          },
          {
            "label": "Criminal",
            "value": "36.8%"
          },
          {
            "label": "Family",
            "value": "9.1%"
          },
          {
            "label": "Health",
            "value": "66.7%"
          },
          {
            "label": "Immigration",
            "value": "36.4%"
          },
          {
            "label": "Weighted Pass",
            "value": "84.4%"
          }
        ]
      },
      {
        "id": "grok-grok-4-6",
        "label": "grok/grok-4.6",
        "model": "grok/grok-4.6",
        "provider": "SpaceXAI",
        "harness": null,
        "effort": "high",
        "score": 48.077,
        "costUsd": 1.530204,
        "uncertainty": {
          "lower": 44.604,
          "upper": 51.55,
          "label": "±1 standard error"
        },
        "sourceUrl": "https://www.vals.ai/benchmarks/legal_research",
        "details": [
          {
            "label": "Publisher model id",
            "value": "grok/grok-4.6"
          },
          {
            "label": "Standard error",
            "value": "±3.47 points"
          },
          {
            "label": "Mean latency",
            "value": "1,504 seconds per test"
          },
          {
            "label": "Evaluation mode",
            "value": "Agentic · many steps allowed"
          },
          {
            "label": "Administrative / Regulatory",
            "value": "62.1%"
          },
          {
            "label": "Business & Commercial",
            "value": "48.8%"
          },
          {
            "label": "Civil Litigation",
            "value": "37.0%"
          },
          {
            "label": "Constitutional / Civil Rights",
            "value": "36.8%"
          },
          {
            "label": "Criminal",
            "value": "42.1%"
          },
          {
            "label": "Family",
            "value": "22.7%"
          },
          {
            "label": "Health",
            "value": "73.3%"
          },
          {
            "label": "Immigration",
            "value": "54.5%"
          },
          {
            "label": "Weighted Pass",
            "value": "88.3%"
          }
        ]
      },
      {
        "id": "grok-grok-4-7",
        "label": "grok/grok-4.7",
        "model": "grok/grok-4.7",
        "provider": "SpaceXAI",
        "harness": null,
        "effort": "xhigh",
        "score": 47.115,
        "costUsd": 4.920812,
        "uncertainty": {
          "lower": 43.646,
          "upper": 50.584,
          "label": "±1 standard error"
        },
        "sourceUrl": "https://www.vals.ai/benchmarks/legal_research",
        "details": [
          {
            "label": "Publisher model id",
            "value": "grok/grok-4.7"
          },
          {
            "label": "Standard error",
            "value": "±3.47 points"
          },
          {
            "label": "Mean latency",
            "value": "1,274 seconds per test"
          },
          {
            "label": "Evaluation mode",
            "value": "Agentic · many steps allowed"
          },
          {
            "label": "Administrative / Regulatory",
            "value": "54.5%"
          },
          {
            "label": "Business & Commercial",
            "value": "44.2%"
          },
          {
            "label": "Civil Litigation",
            "value": "44.4%"
          },
          {
            "label": "Constitutional / Civil Rights",
            "value": "42.1%"
          },
          {
            "label": "Criminal",
            "value": "47.4%"
          },
          {
            "label": "Family",
            "value": "18.2%"
          },
          {
            "label": "Health",
            "value": "76.7%"
          },
          {
            "label": "Immigration",
            "value": "45.5%"
          },
          {
            "label": "Weighted Pass",
            "value": "86.6%"
          }
        ]
      },
      {
        "id": "inception-mercury-2-5",
        "label": "inception/mercury-2.5",
        "model": "inception/mercury-2.5",
        "provider": "Inception",
        "harness": null,
        "effort": "high",
        "score": 4.327,
        "costUsd": 0.058112,
        "uncertainty": {
          "lower": 2.9130000000000003,
          "upper": 5.741,
          "label": "±1 standard error"
        },
        "sourceUrl": "https://www.vals.ai/benchmarks/legal_research",
        "details": [
          {
            "label": "Publisher model id",
            "value": "inception/mercury-2.5"
          },
          {
            "label": "Standard error",
            "value": "±1.41 points"
          },
          {
            "label": "Mean latency",
            "value": "46 seconds per test"
          },
          {
            "label": "Evaluation mode",
            "value": "Agentic · many steps allowed"
          },
          {
            "label": "Administrative / Regulatory",
            "value": "4.5%"
          },
          {
            "label": "Business & Commercial",
            "value": "11.6%"
          },
          {
            "label": "Civil Litigation",
            "value": "3.7%"
          },
          {
            "label": "Constitutional / Civil Rights",
            "value": "2.6%"
          },
          {
            "label": "Criminal",
            "value": "1.8%"
          },
          {
            "label": "Family",
            "value": "0.0%"
          },
          {
            "label": "Health",
            "value": "3.3%"
          },
          {
            "label": "Immigration",
            "value": "0.0%"
          },
          {
            "label": "Weighted Pass",
            "value": "58.9%"
          }
        ]
      },
      {
        "id": "kimi-kimi-k2-5-thinking",
        "label": "kimi/kimi-k2.5-thinking",
        "model": "kimi/kimi-k2.5-thinking",
        "provider": "Moonshot AI",
        "harness": null,
        "effort": null,
        "score": 15.865,
        "costUsd": 0.23035,
        "uncertainty": {
          "lower": 13.326,
          "upper": 18.404,
          "label": "±1 standard error"
        },
        "sourceUrl": "https://www.vals.ai/benchmarks/legal_research",
        "details": [
          {
            "label": "Publisher model id",
            "value": "kimi/kimi-k2.5-thinking"
          },
          {
            "label": "Standard error",
            "value": "±2.54 points"
          },
          {
            "label": "Mean latency",
            "value": "1,333 seconds per test"
          },
          {
            "label": "Evaluation mode",
            "value": "Agentic · many steps allowed"
          },
          {
            "label": "Administrative / Regulatory",
            "value": "16.7%"
          },
          {
            "label": "Business & Commercial",
            "value": "14.0%"
          },
          {
            "label": "Civil Litigation",
            "value": "18.5%"
          },
          {
            "label": "Constitutional / Civil Rights",
            "value": "15.8%"
          },
          {
            "label": "Criminal",
            "value": "15.8%"
          },
          {
            "label": "Family",
            "value": "9.1%"
          },
          {
            "label": "Health",
            "value": "10.0%"
          },
          {
            "label": "Immigration",
            "value": "27.3%"
          },
          {
            "label": "Weighted Pass",
            "value": "71.3%"
          }
        ]
      },
      {
        "id": "kimi-kimi-k2-6",
        "label": "kimi/kimi-k2.6",
        "model": "kimi/kimi-k2.6",
        "provider": "Moonshot AI",
        "harness": null,
        "effort": null,
        "score": 15.865,
        "costUsd": 1.520305,
        "uncertainty": {
          "lower": 13.326,
          "upper": 18.404,
          "label": "±1 standard error"
        },
        "sourceUrl": "https://www.vals.ai/benchmarks/legal_research",
        "details": [
          {
            "label": "Publisher model id",
            "value": "kimi/kimi-k2.6"
          },
          {
            "label": "Standard error",
            "value": "±2.54 points"
          },
          {
            "label": "Mean latency",
            "value": "1,853 seconds per test"
          },
          {
            "label": "Evaluation mode",
            "value": "Agentic · many steps allowed"
          },
          {
            "label": "Administrative / Regulatory",
            "value": "18.2%"
          },
          {
            "label": "Business & Commercial",
            "value": "20.9%"
          },
          {
            "label": "Civil Litigation",
            "value": "18.5%"
          },
          {
            "label": "Constitutional / Civil Rights",
            "value": "21.1%"
          },
          {
            "label": "Criminal",
            "value": "14.0%"
          },
          {
            "label": "Family",
            "value": "0.0%"
          },
          {
            "label": "Health",
            "value": "30.0%"
          },
          {
            "label": "Immigration",
            "value": "18.2%"
          },
          {
            "label": "Weighted Pass",
            "value": "74.7%"
          }
        ]
      },
      {
        "id": "kimi-kimi-k3",
        "label": "kimi/kimi-k3",
        "model": "kimi/kimi-k3",
        "provider": "Moonshot AI",
        "harness": null,
        "effort": null,
        "score": 44.231,
        "costUsd": 3.222748,
        "uncertainty": {
          "lower": 40.779,
          "upper": 47.683,
          "label": "±1 standard error"
        },
        "sourceUrl": "https://www.vals.ai/benchmarks/legal_research",
        "details": [
          {
            "label": "Publisher model id",
            "value": "kimi/kimi-k3"
          },
          {
            "label": "Standard error",
            "value": "±3.45 points"
          },
          {
            "label": "Mean latency",
            "value": "2,146 seconds per test"
          },
          {
            "label": "Evaluation mode",
            "value": "Agentic · many steps allowed"
          },
          {
            "label": "Administrative / Regulatory",
            "value": "54.5%"
          },
          {
            "label": "Business & Commercial",
            "value": "46.5%"
          },
          {
            "label": "Civil Litigation",
            "value": "25.9%"
          },
          {
            "label": "Constitutional / Civil Rights",
            "value": "44.7%"
          },
          {
            "label": "Criminal",
            "value": "36.8%"
          },
          {
            "label": "Family",
            "value": "27.3%"
          },
          {
            "label": "Health",
            "value": "63.3%"
          },
          {
            "label": "Immigration",
            "value": "45.5%"
          },
          {
            "label": "Weighted Pass",
            "value": "87.0%"
          }
        ]
      },
      {
        "id": "meta-muse-spark-1-1",
        "label": "meta/muse_spark_1_1",
        "model": "meta/muse_spark_1_1",
        "provider": "Meta",
        "harness": null,
        "effort": "xhigh",
        "score": 37.981,
        "costUsd": 0.377379,
        "uncertainty": {
          "lower": 34.608000000000004,
          "upper": 41.354,
          "label": "±1 standard error"
        },
        "sourceUrl": "https://www.vals.ai/benchmarks/legal_research",
        "details": [
          {
            "label": "Publisher model id",
            "value": "meta/muse_spark_1_1"
          },
          {
            "label": "Standard error",
            "value": "±3.37 points"
          },
          {
            "label": "Mean latency",
            "value": "313 seconds per test"
          },
          {
            "label": "Evaluation mode",
            "value": "Agentic · many steps allowed"
          },
          {
            "label": "Administrative / Regulatory",
            "value": "37.9%"
          },
          {
            "label": "Business & Commercial",
            "value": "51.2%"
          },
          {
            "label": "Civil Litigation",
            "value": "48.1%"
          },
          {
            "label": "Constitutional / Civil Rights",
            "value": "28.9%"
          },
          {
            "label": "Criminal",
            "value": "36.8%"
          },
          {
            "label": "Family",
            "value": "22.7%"
          },
          {
            "label": "Health",
            "value": "40.0%"
          },
          {
            "label": "Immigration",
            "value": "9.1%"
          },
          {
            "label": "Weighted Pass",
            "value": "84.6%"
          }
        ]
      },
      {
        "id": "meta-muse-spark-1-2",
        "label": "meta/muse_spark_1_2",
        "model": "meta/muse_spark_1_2",
        "provider": "Meta",
        "harness": null,
        "effort": "xhigh",
        "score": 43.75,
        "costUsd": 0.508097,
        "uncertainty": {
          "lower": 40.302,
          "upper": 47.198,
          "label": "±1 standard error"
        },
        "sourceUrl": "https://www.vals.ai/benchmarks/legal_research",
        "details": [
          {
            "label": "Publisher model id",
            "value": "meta/muse_spark_1_2"
          },
          {
            "label": "Standard error",
            "value": "±3.45 points"
          },
          {
            "label": "Mean latency",
            "value": "307 seconds per test"
          },
          {
            "label": "Evaluation mode",
            "value": "Agentic · many steps allowed"
          },
          {
            "label": "Administrative / Regulatory",
            "value": "48.5%"
          },
          {
            "label": "Business & Commercial",
            "value": "44.2%"
          },
          {
            "label": "Civil Litigation",
            "value": "37.0%"
          },
          {
            "label": "Constitutional / Civil Rights",
            "value": "39.5%"
          },
          {
            "label": "Criminal",
            "value": "38.6%"
          },
          {
            "label": "Family",
            "value": "27.3%"
          },
          {
            "label": "Health",
            "value": "56.7%"
          },
          {
            "label": "Immigration",
            "value": "54.5%"
          },
          {
            "label": "Weighted Pass",
            "value": "85.3%"
          }
        ]
      },
      {
        "id": "meta-muse-spark-1-3",
        "label": "meta/muse_spark_1_3",
        "model": "meta/muse_spark_1_3",
        "provider": "Meta",
        "harness": null,
        "effort": "xhigh",
        "score": 40.865,
        "costUsd": 0.650063,
        "uncertainty": {
          "lower": 37.448,
          "upper": 44.282000000000004,
          "label": "±1 standard error"
        },
        "sourceUrl": "https://www.vals.ai/benchmarks/legal_research",
        "details": [
          {
            "label": "Publisher model id",
            "value": "meta/muse_spark_1_3"
          },
          {
            "label": "Standard error",
            "value": "±3.42 points"
          },
          {
            "label": "Mean latency",
            "value": "531 seconds per test"
          },
          {
            "label": "Evaluation mode",
            "value": "Agentic · many steps allowed"
          },
          {
            "label": "Administrative / Regulatory",
            "value": "51.5%"
          },
          {
            "label": "Business & Commercial",
            "value": "32.6%"
          },
          {
            "label": "Civil Litigation",
            "value": "40.7%"
          },
          {
            "label": "Constitutional / Civil Rights",
            "value": "34.2%"
          },
          {
            "label": "Criminal",
            "value": "35.1%"
          },
          {
            "label": "Family",
            "value": "27.3%"
          },
          {
            "label": "Health",
            "value": "70.0%"
          },
          {
            "label": "Immigration",
            "value": "27.3%"
          },
          {
            "label": "Weighted Pass",
            "value": "85.2%"
          }
        ]
      },
      {
        "id": "meta-muse-spark-1-3-max",
        "label": "meta/muse_spark_1_3_max",
        "model": "meta/muse_spark_1_3_max",
        "provider": "Meta",
        "harness": null,
        "effort": "max",
        "score": 55.288,
        "costUsd": 0.588812,
        "uncertainty": {
          "lower": 51.831999999999994,
          "upper": 58.744,
          "label": "±1 standard error"
        },
        "sourceUrl": "https://www.vals.ai/benchmarks/legal_research",
        "details": [
          {
            "label": "Publisher model id",
            "value": "meta/muse_spark_1_3_max"
          },
          {
            "label": "Standard error",
            "value": "±3.46 points"
          },
          {
            "label": "Mean latency",
            "value": "317 seconds per test"
          },
          {
            "label": "Evaluation mode",
            "value": "Agentic · many steps allowed"
          },
          {
            "label": "Administrative / Regulatory",
            "value": "60.6%"
          },
          {
            "label": "Business & Commercial",
            "value": "58.1%"
          },
          {
            "label": "Civil Litigation",
            "value": "63.0%"
          },
          {
            "label": "Constitutional / Civil Rights",
            "value": "50.0%"
          },
          {
            "label": "Criminal",
            "value": "47.4%"
          },
          {
            "label": "Family",
            "value": "40.9%"
          },
          {
            "label": "Health",
            "value": "73.3%"
          },
          {
            "label": "Immigration",
            "value": "45.5%"
          },
          {
            "label": "Weighted Pass",
            "value": "86.8%"
          }
        ]
      },
      {
        "id": "minimax-minimax-m2-7",
        "label": "minimax/MiniMax-M2.7",
        "model": "minimax/MiniMax-M2.7",
        "provider": "MiniMax",
        "harness": null,
        "effort": null,
        "score": 10.577,
        "costUsd": 0.219167,
        "uncertainty": {
          "lower": 8.439,
          "upper": 12.715,
          "label": "±1 standard error"
        },
        "sourceUrl": "https://www.vals.ai/benchmarks/legal_research",
        "details": [
          {
            "label": "Publisher model id",
            "value": "minimax/MiniMax-M2.7"
          },
          {
            "label": "Standard error",
            "value": "±2.14 points"
          },
          {
            "label": "Mean latency",
            "value": "421 seconds per test"
          },
          {
            "label": "Evaluation mode",
            "value": "Agentic · many steps allowed"
          },
          {
            "label": "Administrative / Regulatory",
            "value": "16.7%"
          },
          {
            "label": "Business & Commercial",
            "value": "14.0%"
          },
          {
            "label": "Civil Litigation",
            "value": "14.8%"
          },
          {
            "label": "Constitutional / Civil Rights",
            "value": "10.5%"
          },
          {
            "label": "Criminal",
            "value": "5.3%"
          },
          {
            "label": "Family",
            "value": "0.0%"
          },
          {
            "label": "Health",
            "value": "13.3%"
          },
          {
            "label": "Immigration",
            "value": "0.0%"
          },
          {
            "label": "Weighted Pass",
            "value": "63.9%"
          }
        ]
      },
      {
        "id": "minimax-minimax-m3",
        "label": "minimax/MiniMax-M3",
        "model": "minimax/MiniMax-M3",
        "provider": "MiniMax",
        "harness": null,
        "effort": null,
        "score": 29.808,
        "costUsd": 0.342955,
        "uncertainty": {
          "lower": 26.629,
          "upper": 32.987,
          "label": "±1 standard error"
        },
        "sourceUrl": "https://www.vals.ai/benchmarks/legal_research",
        "details": [
          {
            "label": "Publisher model id",
            "value": "minimax/MiniMax-M3"
          },
          {
            "label": "Standard error",
            "value": "±3.18 points"
          },
          {
            "label": "Mean latency",
            "value": "814 seconds per test"
          },
          {
            "label": "Evaluation mode",
            "value": "Agentic · many steps allowed"
          },
          {
            "label": "Administrative / Regulatory",
            "value": "36.4%"
          },
          {
            "label": "Business & Commercial",
            "value": "25.6%"
          },
          {
            "label": "Civil Litigation",
            "value": "25.9%"
          },
          {
            "label": "Constitutional / Civil Rights",
            "value": "34.2%"
          },
          {
            "label": "Criminal",
            "value": "26.3%"
          },
          {
            "label": "Family",
            "value": "13.6%"
          },
          {
            "label": "Health",
            "value": "46.7%"
          },
          {
            "label": "Immigration",
            "value": "27.3%"
          },
          {
            "label": "Weighted Pass",
            "value": "81.1%"
          }
        ]
      },
      {
        "id": "mistralai-mistral-medium-3-5",
        "label": "mistralai/mistral-medium-3.5",
        "model": "mistralai/mistral-medium-3.5",
        "provider": "Mistral AI",
        "harness": null,
        "effort": "high",
        "score": 9.135,
        "costUsd": 1.308079,
        "uncertainty": {
          "lower": 7.133,
          "upper": 11.137,
          "label": "±1 standard error"
        },
        "sourceUrl": "https://www.vals.ai/benchmarks/legal_research",
        "details": [
          {
            "label": "Publisher model id",
            "value": "mistralai/mistral-medium-3.5"
          },
          {
            "label": "Standard error",
            "value": "±2.00 points"
          },
          {
            "label": "Mean latency",
            "value": "613 seconds per test"
          },
          {
            "label": "Evaluation mode",
            "value": "Agentic · many steps allowed"
          },
          {
            "label": "Administrative / Regulatory",
            "value": "7.6%"
          },
          {
            "label": "Business & Commercial",
            "value": "9.3%"
          },
          {
            "label": "Civil Litigation",
            "value": "11.1%"
          },
          {
            "label": "Constitutional / Civil Rights",
            "value": "10.5%"
          },
          {
            "label": "Criminal",
            "value": "10.5%"
          },
          {
            "label": "Family",
            "value": "0.0%"
          },
          {
            "label": "Health",
            "value": "6.7%"
          },
          {
            "label": "Immigration",
            "value": "9.1%"
          },
          {
            "label": "Weighted Pass",
            "value": "63.3%"
          }
        ]
      },
      {
        "id": "nvidia-nemotron-3-ultra-550b-a55b",
        "label": "nvidia/nemotron-3-ultra-550b-a55b",
        "model": "nvidia/nemotron-3-ultra-550b-a55b",
        "provider": "Nvidia",
        "harness": null,
        "effort": null,
        "score": 15.385,
        "costUsd": null,
        "uncertainty": {
          "lower": 12.876999999999999,
          "upper": 17.893,
          "label": "±1 standard error"
        },
        "sourceUrl": "https://www.vals.ai/benchmarks/legal_research",
        "details": [
          {
            "label": "Publisher model id",
            "value": "nvidia/nemotron-3-ultra-550b-a55b"
          },
          {
            "label": "Standard error",
            "value": "±2.51 points"
          },
          {
            "label": "Mean latency",
            "value": "611 seconds per test"
          },
          {
            "label": "Evaluation mode",
            "value": "Agentic · many steps allowed"
          },
          {
            "label": "Administrative / Regulatory",
            "value": "16.7%"
          },
          {
            "label": "Business & Commercial",
            "value": "11.6%"
          },
          {
            "label": "Civil Litigation",
            "value": "22.2%"
          },
          {
            "label": "Constitutional / Civil Rights",
            "value": "13.2%"
          },
          {
            "label": "Criminal",
            "value": "15.8%"
          },
          {
            "label": "Family",
            "value": "4.5%"
          },
          {
            "label": "Health",
            "value": "16.7%"
          },
          {
            "label": "Immigration",
            "value": "9.1%"
          },
          {
            "label": "Weighted Pass",
            "value": "68.0%"
          }
        ]
      },
      {
        "id": "openai-gpt-5-4-mini-2026-03-17",
        "label": "openai/gpt-5.4-mini-2026-03-17",
        "model": "openai/gpt-5.4-mini-2026-03-17",
        "provider": "OpenAI",
        "harness": null,
        "effort": "xhigh",
        "score": 12.5,
        "costUsd": 1.864673,
        "uncertainty": {
          "lower": 10.201,
          "upper": 14.799,
          "label": "±1 standard error"
        },
        "sourceUrl": "https://www.vals.ai/benchmarks/legal_research",
        "details": [
          {
            "label": "Publisher model id",
            "value": "openai/gpt-5.4-mini-2026-03-17"
          },
          {
            "label": "Standard error",
            "value": "±2.30 points"
          },
          {
            "label": "Mean latency",
            "value": "2,256 seconds per test"
          },
          {
            "label": "Evaluation mode",
            "value": "Agentic · many steps allowed"
          },
          {
            "label": "Administrative / Regulatory",
            "value": "15.2%"
          },
          {
            "label": "Business & Commercial",
            "value": "11.6%"
          },
          {
            "label": "Civil Litigation",
            "value": "14.8%"
          },
          {
            "label": "Constitutional / Civil Rights",
            "value": "13.2%"
          },
          {
            "label": "Criminal",
            "value": "5.3%"
          },
          {
            "label": "Family",
            "value": "0.0%"
          },
          {
            "label": "Health",
            "value": "16.7%"
          },
          {
            "label": "Immigration",
            "value": "18.2%"
          },
          {
            "label": "Weighted Pass",
            "value": "67.3%"
          }
        ]
      },
      {
        "id": "openai-gpt-5-4-nano-2026-03-17",
        "label": "openai/gpt-5.4-nano-2026-03-17",
        "model": "openai/gpt-5.4-nano-2026-03-17",
        "provider": "OpenAI",
        "harness": null,
        "effort": "high",
        "score": 6.25,
        "costUsd": 0.141253,
        "uncertainty": {
          "lower": 4.568,
          "upper": 7.932,
          "label": "±1 standard error"
        },
        "sourceUrl": "https://www.vals.ai/benchmarks/legal_research",
        "details": [
          {
            "label": "Publisher model id",
            "value": "openai/gpt-5.4-nano-2026-03-17"
          },
          {
            "label": "Standard error",
            "value": "±1.68 points"
          },
          {
            "label": "Mean latency",
            "value": "527 seconds per test"
          },
          {
            "label": "Evaluation mode",
            "value": "Agentic · many steps allowed"
          },
          {
            "label": "Administrative / Regulatory",
            "value": "10.6%"
          },
          {
            "label": "Business & Commercial",
            "value": "7.0%"
          },
          {
            "label": "Civil Litigation",
            "value": "7.4%"
          },
          {
            "label": "Constitutional / Civil Rights",
            "value": "2.6%"
          },
          {
            "label": "Criminal",
            "value": "3.5%"
          },
          {
            "label": "Family",
            "value": "0.0%"
          },
          {
            "label": "Health",
            "value": "13.3%"
          },
          {
            "label": "Immigration",
            "value": "0.0%"
          },
          {
            "label": "Weighted Pass",
            "value": "62.0%"
          }
        ]
      },
      {
        "id": "openai-gpt-5-5",
        "label": "openai/gpt-5.5",
        "model": "openai/gpt-5.5",
        "provider": "OpenAI",
        "harness": null,
        "effort": "xhigh",
        "score": 40.385,
        "costUsd": 7.402471,
        "uncertainty": {
          "lower": 36.974999999999994,
          "upper": 43.795,
          "label": "±1 standard error"
        },
        "sourceUrl": "https://www.vals.ai/benchmarks/legal_research",
        "details": [
          {
            "label": "Publisher model id",
            "value": "openai/gpt-5.5"
          },
          {
            "label": "Standard error",
            "value": "±3.41 points"
          },
          {
            "label": "Mean latency",
            "value": "2,265 seconds per test"
          },
          {
            "label": "Evaluation mode",
            "value": "Agentic · many steps allowed"
          },
          {
            "label": "Administrative / Regulatory",
            "value": "53.0%"
          },
          {
            "label": "Business & Commercial",
            "value": "37.2%"
          },
          {
            "label": "Civil Litigation",
            "value": "44.4%"
          },
          {
            "label": "Constitutional / Civil Rights",
            "value": "28.9%"
          },
          {
            "label": "Criminal",
            "value": "36.8%"
          },
          {
            "label": "Family",
            "value": "13.6%"
          },
          {
            "label": "Health",
            "value": "66.7%"
          },
          {
            "label": "Immigration",
            "value": "36.4%"
          },
          {
            "label": "Weighted Pass",
            "value": "84.0%"
          }
        ]
      },
      {
        "id": "openai-gpt-5-6-luna",
        "label": "openai/gpt-5.6-luna",
        "model": "openai/gpt-5.6-luna",
        "provider": "OpenAI",
        "harness": null,
        "effort": "max",
        "score": 36.538,
        "costUsd": 0.845008,
        "uncertainty": {
          "lower": 33.190999999999995,
          "upper": 39.885,
          "label": "±1 standard error"
        },
        "sourceUrl": "https://www.vals.ai/benchmarks/legal_research",
        "details": [
          {
            "label": "Publisher model id",
            "value": "openai/gpt-5.6-luna"
          },
          {
            "label": "Standard error",
            "value": "±3.35 points"
          },
          {
            "label": "Mean latency",
            "value": "2,374 seconds per test"
          },
          {
            "label": "Evaluation mode",
            "value": "Agentic · many steps allowed"
          },
          {
            "label": "Administrative / Regulatory",
            "value": "48.5%"
          },
          {
            "label": "Business & Commercial",
            "value": "32.6%"
          },
          {
            "label": "Civil Litigation",
            "value": "25.9%"
          },
          {
            "label": "Constitutional / Civil Rights",
            "value": "34.2%"
          },
          {
            "label": "Criminal",
            "value": "29.8%"
          },
          {
            "label": "Family",
            "value": "18.2%"
          },
          {
            "label": "Health",
            "value": "70.0%"
          },
          {
            "label": "Immigration",
            "value": "18.2%"
          },
          {
            "label": "Weighted Pass",
            "value": "85.0%"
          }
        ]
      },
      {
        "id": "openai-gpt-5-6-sol",
        "label": "openai/gpt-5.6-sol",
        "model": "openai/gpt-5.6-sol",
        "provider": "OpenAI",
        "harness": null,
        "effort": "max",
        "score": 48.077,
        "costUsd": 21.605566,
        "uncertainty": {
          "lower": 44.604,
          "upper": 51.55,
          "label": "±1 standard error"
        },
        "sourceUrl": "https://www.vals.ai/benchmarks/legal_research",
        "details": [
          {
            "label": "Publisher model id",
            "value": "openai/gpt-5.6-sol"
          },
          {
            "label": "Standard error",
            "value": "±3.47 points"
          },
          {
            "label": "Mean latency",
            "value": "4,631 seconds per test"
          },
          {
            "label": "Evaluation mode",
            "value": "Agentic · many steps allowed"
          },
          {
            "label": "Administrative / Regulatory",
            "value": "60.6%"
          },
          {
            "label": "Business & Commercial",
            "value": "53.5%"
          },
          {
            "label": "Civil Litigation",
            "value": "40.7%"
          },
          {
            "label": "Constitutional / Civil Rights",
            "value": "34.2%"
          },
          {
            "label": "Criminal",
            "value": "43.9%"
          },
          {
            "label": "Family",
            "value": "13.6%"
          },
          {
            "label": "Health",
            "value": "80.0%"
          },
          {
            "label": "Immigration",
            "value": "45.5%"
          },
          {
            "label": "Weighted Pass",
            "value": "87.8%"
          }
        ]
      },
      {
        "id": "openai-gpt-5-6-terra",
        "label": "openai/gpt-5.6-terra",
        "model": "openai/gpt-5.6-terra",
        "provider": "OpenAI",
        "harness": null,
        "effort": "max",
        "score": 41.346,
        "costUsd": 7.907299,
        "uncertainty": {
          "lower": 37.922999999999995,
          "upper": 44.769,
          "label": "±1 standard error"
        },
        "sourceUrl": "https://www.vals.ai/benchmarks/legal_research",
        "details": [
          {
            "label": "Publisher model id",
            "value": "openai/gpt-5.6-terra"
          },
          {
            "label": "Standard error",
            "value": "±3.42 points"
          },
          {
            "label": "Mean latency",
            "value": "4,173 seconds per test"
          },
          {
            "label": "Evaluation mode",
            "value": "Agentic · many steps allowed"
          },
          {
            "label": "Administrative / Regulatory",
            "value": "57.6%"
          },
          {
            "label": "Business & Commercial",
            "value": "37.2%"
          },
          {
            "label": "Civil Litigation",
            "value": "29.6%"
          },
          {
            "label": "Constitutional / Civil Rights",
            "value": "34.2%"
          },
          {
            "label": "Criminal",
            "value": "35.1%"
          },
          {
            "label": "Family",
            "value": "18.2%"
          },
          {
            "label": "Health",
            "value": "80.0%"
          },
          {
            "label": "Immigration",
            "value": "18.2%"
          },
          {
            "label": "Weighted Pass",
            "value": "85.1%"
          }
        ]
      },
      {
        "id": "openai-gpt-6-astra",
        "label": "openai/gpt-6-astra",
        "model": "openai/gpt-6-astra",
        "provider": "OpenAI",
        "harness": null,
        "effort": "max",
        "score": 39.423,
        "costUsd": 10.577143,
        "uncertainty": {
          "lower": 36.026,
          "upper": 42.82,
          "label": "±1 standard error"
        },
        "sourceUrl": "https://www.vals.ai/benchmarks/legal_research",
        "details": [
          {
            "label": "Publisher model id",
            "value": "openai/gpt-6-astra"
          },
          {
            "label": "Standard error",
            "value": "±3.40 points"
          },
          {
            "label": "Mean latency",
            "value": "1,491 seconds per test"
          },
          {
            "label": "Evaluation mode",
            "value": "Agentic · many steps allowed"
          },
          {
            "label": "Administrative / Regulatory",
            "value": "53.0%"
          },
          {
            "label": "Business & Commercial",
            "value": "32.6%"
          },
          {
            "label": "Civil Litigation",
            "value": "40.7%"
          },
          {
            "label": "Constitutional / Civil Rights",
            "value": "26.3%"
          },
          {
            "label": "Criminal",
            "value": "29.8%"
          },
          {
            "label": "Family",
            "value": "18.2%"
          },
          {
            "label": "Health",
            "value": "76.7%"
          },
          {
            "label": "Immigration",
            "value": "36.4%"
          },
          {
            "label": "Weighted Pass",
            "value": "84.8%"
          }
        ]
      },
      {
        "id": "openai-gpt-6-luna",
        "label": "openai/gpt-6-luna",
        "model": "openai/gpt-6-luna",
        "provider": "OpenAI",
        "harness": null,
        "effort": "max",
        "score": 30.288,
        "costUsd": 0.440296,
        "uncertainty": {
          "lower": 27.094,
          "upper": 33.482,
          "label": "±1 standard error"
        },
        "sourceUrl": "https://www.vals.ai/benchmarks/legal_research",
        "details": [
          {
            "label": "Publisher model id",
            "value": "openai/gpt-6-luna"
          },
          {
            "label": "Standard error",
            "value": "±3.19 points"
          },
          {
            "label": "Mean latency",
            "value": "2,548 seconds per test"
          },
          {
            "label": "Evaluation mode",
            "value": "Agentic · many steps allowed"
          },
          {
            "label": "Administrative / Regulatory",
            "value": "37.9%"
          },
          {
            "label": "Business & Commercial",
            "value": "23.3%"
          },
          {
            "label": "Civil Litigation",
            "value": "25.9%"
          },
          {
            "label": "Constitutional / Civil Rights",
            "value": "31.6%"
          },
          {
            "label": "Criminal",
            "value": "26.3%"
          },
          {
            "label": "Family",
            "value": "9.1%"
          },
          {
            "label": "Health",
            "value": "53.3%"
          },
          {
            "label": "Immigration",
            "value": "45.5%"
          },
          {
            "label": "Weighted Pass",
            "value": "79.3%"
          }
        ]
      },
      {
        "id": "openai-gpt-6-sol",
        "label": "openai/gpt-6-sol",
        "model": "openai/gpt-6-sol",
        "provider": "OpenAI",
        "harness": null,
        "effort": "max",
        "score": 28.846,
        "costUsd": 4.897717,
        "uncertainty": {
          "lower": 25.697,
          "upper": 31.995,
          "label": "±1 standard error"
        },
        "sourceUrl": "https://www.vals.ai/benchmarks/legal_research",
        "details": [
          {
            "label": "Publisher model id",
            "value": "openai/gpt-6-sol"
          },
          {
            "label": "Standard error",
            "value": "±3.15 points"
          },
          {
            "label": "Mean latency",
            "value": "1,449 seconds per test"
          },
          {
            "label": "Evaluation mode",
            "value": "Agentic · many steps allowed"
          },
          {
            "label": "Administrative / Regulatory",
            "value": "39.4%"
          },
          {
            "label": "Business & Commercial",
            "value": "27.9%"
          },
          {
            "label": "Civil Litigation",
            "value": "18.5%"
          },
          {
            "label": "Constitutional / Civil Rights",
            "value": "18.4%"
          },
          {
            "label": "Criminal",
            "value": "26.3%"
          },
          {
            "label": "Family",
            "value": "9.1%"
          },
          {
            "label": "Health",
            "value": "60.0%"
          },
          {
            "label": "Immigration",
            "value": "36.4%"
          },
          {
            "label": "Weighted Pass",
            "value": "79.2%"
          }
        ]
      },
      {
        "id": "poolside-laguna-m-1",
        "label": "poolside/laguna-m.1",
        "model": "poolside/laguna-m.1",
        "provider": "Poolside",
        "harness": null,
        "effort": null,
        "score": 2.404,
        "costUsd": 0.219839,
        "uncertainty": {
          "lower": 1.339,
          "upper": 3.469,
          "label": "±1 standard error"
        },
        "sourceUrl": "https://www.vals.ai/benchmarks/legal_research",
        "details": [
          {
            "label": "Publisher model id",
            "value": "poolside/laguna-m.1"
          },
          {
            "label": "Standard error",
            "value": "±1.06 points"
          },
          {
            "label": "Mean latency",
            "value": "466 seconds per test"
          },
          {
            "label": "Evaluation mode",
            "value": "Agentic · many steps allowed"
          },
          {
            "label": "Administrative / Regulatory",
            "value": "3.0%"
          },
          {
            "label": "Business & Commercial",
            "value": "4.7%"
          },
          {
            "label": "Civil Litigation",
            "value": "0.0%"
          },
          {
            "label": "Constitutional / Civil Rights",
            "value": "2.6%"
          },
          {
            "label": "Criminal",
            "value": "0.0%"
          },
          {
            "label": "Family",
            "value": "0.0%"
          },
          {
            "label": "Health",
            "value": "3.3%"
          },
          {
            "label": "Immigration",
            "value": "0.0%"
          },
          {
            "label": "Weighted Pass",
            "value": "62.3%"
          }
        ]
      },
      {
        "id": "poolside-laguna-xs-2",
        "label": "poolside/laguna-xs.2",
        "model": "poolside/laguna-xs.2",
        "provider": "Poolside",
        "harness": null,
        "effort": null,
        "score": 0.962,
        "costUsd": 0.107892,
        "uncertainty": {
          "lower": 0.2839999999999999,
          "upper": 1.6400000000000001,
          "label": "±1 standard error"
        },
        "sourceUrl": "https://www.vals.ai/benchmarks/legal_research",
        "details": [
          {
            "label": "Publisher model id",
            "value": "poolside/laguna-xs.2"
          },
          {
            "label": "Standard error",
            "value": "±0.68 points"
          },
          {
            "label": "Mean latency",
            "value": "352 seconds per test"
          },
          {
            "label": "Evaluation mode",
            "value": "Agentic · many steps allowed"
          },
          {
            "label": "Administrative / Regulatory",
            "value": "1.5%"
          },
          {
            "label": "Business & Commercial",
            "value": "0.0%"
          },
          {
            "label": "Civil Litigation",
            "value": "3.7%"
          },
          {
            "label": "Constitutional / Civil Rights",
            "value": "0.0%"
          },
          {
            "label": "Criminal",
            "value": "0.0%"
          },
          {
            "label": "Family",
            "value": "0.0%"
          },
          {
            "label": "Health",
            "value": "0.0%"
          },
          {
            "label": "Immigration",
            "value": "0.0%"
          },
          {
            "label": "Weighted Pass",
            "value": "56.3%"
          }
        ]
      },
      {
        "id": "tencent-hy4-preview",
        "label": "tencent/hy4-preview",
        "model": "tencent/hy4-preview",
        "provider": "Tencent",
        "harness": null,
        "effort": null,
        "score": 45.192,
        "costUsd": 0.846372,
        "uncertainty": {
          "lower": 41.733,
          "upper": 48.651,
          "label": "±1 standard error"
        },
        "sourceUrl": "https://www.vals.ai/benchmarks/legal_research",
        "details": [
          {
            "label": "Publisher model id",
            "value": "tencent/hy4-preview"
          },
          {
            "label": "Standard error",
            "value": "±3.46 points"
          },
          {
            "label": "Mean latency",
            "value": "3,840 seconds per test"
          },
          {
            "label": "Evaluation mode",
            "value": "Agentic · many steps allowed"
          },
          {
            "label": "Administrative / Regulatory",
            "value": "54.5%"
          },
          {
            "label": "Business & Commercial",
            "value": "44.2%"
          },
          {
            "label": "Civil Litigation",
            "value": "59.3%"
          },
          {
            "label": "Constitutional / Civil Rights",
            "value": "39.5%"
          },
          {
            "label": "Criminal",
            "value": "36.8%"
          },
          {
            "label": "Family",
            "value": "31.8%"
          },
          {
            "label": "Health",
            "value": "66.7%"
          },
          {
            "label": "Immigration",
            "value": "27.3%"
          },
          {
            "label": "Weighted Pass",
            "value": "85.2%"
          }
        ]
      },
      {
        "id": "thinkingmachines-inkling",
        "label": "thinkingmachines/inkling",
        "model": "thinkingmachines/inkling",
        "provider": "Thinkingmachines",
        "harness": null,
        "effort": "0.99",
        "score": 28.365,
        "costUsd": 0.494044,
        "uncertainty": {
          "lower": 25.232,
          "upper": 31.497999999999998,
          "label": "±1 standard error"
        },
        "sourceUrl": "https://www.vals.ai/benchmarks/legal_research",
        "details": [
          {
            "label": "Publisher model id",
            "value": "thinkingmachines/inkling"
          },
          {
            "label": "Standard error",
            "value": "±3.13 points"
          },
          {
            "label": "Mean latency",
            "value": "869 seconds per test"
          },
          {
            "label": "Evaluation mode",
            "value": "Agentic · many steps allowed"
          },
          {
            "label": "Administrative / Regulatory",
            "value": "31.8%"
          },
          {
            "label": "Business & Commercial",
            "value": "25.6%"
          },
          {
            "label": "Civil Litigation",
            "value": "33.3%"
          },
          {
            "label": "Constitutional / Civil Rights",
            "value": "31.6%"
          },
          {
            "label": "Criminal",
            "value": "21.1%"
          },
          {
            "label": "Family",
            "value": "13.6%"
          },
          {
            "label": "Health",
            "value": "43.3%"
          },
          {
            "label": "Immigration",
            "value": "36.4%"
          },
          {
            "label": "Weighted Pass",
            "value": "80.2%"
          }
        ]
      },
      {
        "id": "thinkingmachines-inkling-small",
        "label": "thinkingmachines/inkling-small",
        "model": "thinkingmachines/inkling-small",
        "provider": "Thinkingmachines",
        "harness": null,
        "effort": "0.99",
        "score": 25.481,
        "costUsd": 0.141548,
        "uncertainty": {
          "lower": 22.452,
          "upper": 28.51,
          "label": "±1 standard error"
        },
        "sourceUrl": "https://www.vals.ai/benchmarks/legal_research",
        "details": [
          {
            "label": "Publisher model id",
            "value": "thinkingmachines/inkling-small"
          },
          {
            "label": "Standard error",
            "value": "±3.03 points"
          },
          {
            "label": "Mean latency",
            "value": "775 seconds per test"
          },
          {
            "label": "Evaluation mode",
            "value": "Agentic · many steps allowed"
          },
          {
            "label": "Administrative / Regulatory",
            "value": "28.8%"
          },
          {
            "label": "Business & Commercial",
            "value": "34.9%"
          },
          {
            "label": "Civil Litigation",
            "value": "22.2%"
          },
          {
            "label": "Constitutional / Civil Rights",
            "value": "10.5%"
          },
          {
            "label": "Criminal",
            "value": "24.6%"
          },
          {
            "label": "Family",
            "value": "4.5%"
          },
          {
            "label": "Health",
            "value": "30.0%"
          },
          {
            "label": "Immigration",
            "value": "18.2%"
          },
          {
            "label": "Weighted Pass",
            "value": "78.9%"
          }
        ]
      },
      {
        "id": "xiaomi-mimo-v2-5",
        "label": "xiaomi/mimo-v2.5",
        "model": "xiaomi/mimo-v2.5",
        "provider": "Xiaomi",
        "harness": null,
        "effort": null,
        "score": 9.135,
        "costUsd": 0.033318,
        "uncertainty": {
          "lower": 7.133,
          "upper": 11.137,
          "label": "±1 standard error"
        },
        "sourceUrl": "https://www.vals.ai/benchmarks/legal_research",
        "details": [
          {
            "label": "Publisher model id",
            "value": "xiaomi/mimo-v2.5"
          },
          {
            "label": "Standard error",
            "value": "±2.00 points"
          },
          {
            "label": "Mean latency",
            "value": "391 seconds per test"
          },
          {
            "label": "Evaluation mode",
            "value": "Agentic · many steps allowed"
          },
          {
            "label": "Administrative / Regulatory",
            "value": "10.6%"
          },
          {
            "label": "Business & Commercial",
            "value": "9.3%"
          },
          {
            "label": "Civil Litigation",
            "value": "11.1%"
          },
          {
            "label": "Constitutional / Civil Rights",
            "value": "7.9%"
          },
          {
            "label": "Criminal",
            "value": "7.0%"
          },
          {
            "label": "Family",
            "value": "0.0%"
          },
          {
            "label": "Health",
            "value": "6.7%"
          },
          {
            "label": "Immigration",
            "value": "9.1%"
          },
          {
            "label": "Weighted Pass",
            "value": "68.2%"
          }
        ]
      },
      {
        "id": "xiaomi-mimo-v2-5-pro",
        "label": "xiaomi/mimo-v2.5-pro",
        "model": "xiaomi/mimo-v2.5-pro",
        "provider": "Xiaomi",
        "harness": null,
        "effort": null,
        "score": 15.865,
        "costUsd": 0.082648,
        "uncertainty": {
          "lower": 13.326,
          "upper": 18.404,
          "label": "±1 standard error"
        },
        "sourceUrl": "https://www.vals.ai/benchmarks/legal_research",
        "details": [
          {
            "label": "Publisher model id",
            "value": "xiaomi/mimo-v2.5-pro"
          },
          {
            "label": "Standard error",
            "value": "±2.54 points"
          },
          {
            "label": "Mean latency",
            "value": "416 seconds per test"
          },
          {
            "label": "Evaluation mode",
            "value": "Agentic · many steps allowed"
          },
          {
            "label": "Administrative / Regulatory",
            "value": "18.2%"
          },
          {
            "label": "Business & Commercial",
            "value": "11.6%"
          },
          {
            "label": "Civil Litigation",
            "value": "11.1%"
          },
          {
            "label": "Constitutional / Civil Rights",
            "value": "13.2%"
          },
          {
            "label": "Criminal",
            "value": "14.0%"
          },
          {
            "label": "Family",
            "value": "4.5%"
          },
          {
            "label": "Health",
            "value": "23.3%"
          },
          {
            "label": "Immigration",
            "value": "27.3%"
          },
          {
            "label": "Weighted Pass",
            "value": "76.0%"
          }
        ]
      },
      {
        "id": "xiaomi-mimo-v2-6-flash",
        "label": "xiaomi/mimo-v2.6-flash",
        "model": "xiaomi/mimo-v2.6-flash",
        "provider": "Xiaomi",
        "harness": null,
        "effort": null,
        "score": 37.981,
        "costUsd": 0.070424,
        "uncertainty": {
          "lower": 34.608000000000004,
          "upper": 41.354,
          "label": "±1 standard error"
        },
        "sourceUrl": "https://www.vals.ai/benchmarks/legal_research",
        "details": [
          {
            "label": "Publisher model id",
            "value": "xiaomi/mimo-v2.6-flash"
          },
          {
            "label": "Standard error",
            "value": "±3.37 points"
          },
          {
            "label": "Mean latency",
            "value": "920 seconds per test"
          },
          {
            "label": "Evaluation mode",
            "value": "Agentic · many steps allowed"
          },
          {
            "label": "Administrative / Regulatory",
            "value": "47.0%"
          },
          {
            "label": "Business & Commercial",
            "value": "37.2%"
          },
          {
            "label": "Civil Litigation",
            "value": "29.6%"
          },
          {
            "label": "Constitutional / Civil Rights",
            "value": "34.2%"
          },
          {
            "label": "Criminal",
            "value": "29.8%"
          },
          {
            "label": "Family",
            "value": "31.8%"
          },
          {
            "label": "Health",
            "value": "53.3%"
          },
          {
            "label": "Immigration",
            "value": "18.2%"
          },
          {
            "label": "Weighted Pass",
            "value": "79.6%"
          }
        ]
      },
      {
        "id": "xiaomi-mimo-v2-6-pro",
        "label": "xiaomi/mimo-v2.6-pro",
        "model": "xiaomi/mimo-v2.6-pro",
        "provider": "Xiaomi",
        "harness": null,
        "effort": null,
        "score": 47.115,
        "costUsd": 0.181201,
        "uncertainty": {
          "lower": 43.646,
          "upper": 50.584,
          "label": "±1 standard error"
        },
        "sourceUrl": "https://www.vals.ai/benchmarks/legal_research",
        "details": [
          {
            "label": "Publisher model id",
            "value": "xiaomi/mimo-v2.6-pro"
          },
          {
            "label": "Standard error",
            "value": "±3.47 points"
          },
          {
            "label": "Mean latency",
            "value": "1,816 seconds per test"
          },
          {
            "label": "Evaluation mode",
            "value": "Agentic · many steps allowed"
          },
          {
            "label": "Administrative / Regulatory",
            "value": "59.1%"
          },
          {
            "label": "Business & Commercial",
            "value": "44.2%"
          },
          {
            "label": "Civil Litigation",
            "value": "40.7%"
          },
          {
            "label": "Constitutional / Civil Rights",
            "value": "47.4%"
          },
          {
            "label": "Criminal",
            "value": "43.9%"
          },
          {
            "label": "Family",
            "value": "27.3%"
          },
          {
            "label": "Health",
            "value": "70.0%"
          },
          {
            "label": "Immigration",
            "value": "27.3%"
          },
          {
            "label": "Weighted Pass",
            "value": "85.5%"
          }
        ]
      },
      {
        "id": "zai-glm-5-1",
        "label": "zai/glm-5.1",
        "model": "zai/glm-5.1",
        "provider": "Zhipu AI",
        "harness": null,
        "effort": null,
        "score": 27.885,
        "costUsd": 0.385737,
        "uncertainty": {
          "lower": 24.768,
          "upper": 31.002000000000002,
          "label": "±1 standard error"
        },
        "sourceUrl": "https://www.vals.ai/benchmarks/legal_research",
        "details": [
          {
            "label": "Publisher model id",
            "value": "zai/glm-5.1"
          },
          {
            "label": "Standard error",
            "value": "±3.12 points"
          },
          {
            "label": "Mean latency",
            "value": "326 seconds per test"
          },
          {
            "label": "Evaluation mode",
            "value": "Agentic · many steps allowed"
          },
          {
            "label": "Administrative / Regulatory",
            "value": "31.8%"
          },
          {
            "label": "Business & Commercial",
            "value": "23.3%"
          },
          {
            "label": "Civil Litigation",
            "value": "22.2%"
          },
          {
            "label": "Constitutional / Civil Rights",
            "value": "34.2%"
          },
          {
            "label": "Criminal",
            "value": "29.8%"
          },
          {
            "label": "Family",
            "value": "9.1%"
          },
          {
            "label": "Health",
            "value": "36.7%"
          },
          {
            "label": "Immigration",
            "value": "18.2%"
          },
          {
            "label": "Weighted Pass",
            "value": "77.8%"
          }
        ]
      },
      {
        "id": "zai-glm-5-2",
        "label": "zai/glm-5.2",
        "model": "zai/glm-5.2",
        "provider": "Zhipu AI",
        "harness": null,
        "effort": "max",
        "score": 31.25,
        "costUsd": 0.889209,
        "uncertainty": {
          "lower": 28.028,
          "upper": 34.472,
          "label": "±1 standard error"
        },
        "sourceUrl": "https://www.vals.ai/benchmarks/legal_research",
        "details": [
          {
            "label": "Publisher model id",
            "value": "zai/glm-5.2"
          },
          {
            "label": "Standard error",
            "value": "±3.22 points"
          },
          {
            "label": "Mean latency",
            "value": "1,023 seconds per test"
          },
          {
            "label": "Evaluation mode",
            "value": "Agentic · many steps allowed"
          },
          {
            "label": "Administrative / Regulatory",
            "value": "39.4%"
          },
          {
            "label": "Business & Commercial",
            "value": "27.9%"
          },
          {
            "label": "Civil Litigation",
            "value": "25.9%"
          },
          {
            "label": "Constitutional / Civil Rights",
            "value": "28.9%"
          },
          {
            "label": "Criminal",
            "value": "22.8%"
          },
          {
            "label": "Family",
            "value": "22.7%"
          },
          {
            "label": "Health",
            "value": "50.0%"
          },
          {
            "label": "Immigration",
            "value": "27.3%"
          },
          {
            "label": "Weighted Pass",
            "value": "82.9%"
          }
        ]
      },
      {
        "id": "zai-glm-5-3",
        "label": "zai/glm-5.3",
        "model": "zai/glm-5.3",
        "provider": "Zhipu AI",
        "harness": null,
        "effort": "max",
        "score": 49.038,
        "costUsd": 2.241388,
        "uncertainty": {
          "lower": 45.562999999999995,
          "upper": 52.513,
          "label": "±1 standard error"
        },
        "sourceUrl": "https://www.vals.ai/benchmarks/legal_research",
        "details": [
          {
            "label": "Publisher model id",
            "value": "zai/glm-5.3"
          },
          {
            "label": "Standard error",
            "value": "±3.48 points"
          },
          {
            "label": "Mean latency",
            "value": "3,110 seconds per test"
          },
          {
            "label": "Evaluation mode",
            "value": "Agentic · many steps allowed"
          },
          {
            "label": "Administrative / Regulatory",
            "value": "62.1%"
          },
          {
            "label": "Business & Commercial",
            "value": "46.5%"
          },
          {
            "label": "Civil Litigation",
            "value": "55.6%"
          },
          {
            "label": "Constitutional / Civil Rights",
            "value": "39.5%"
          },
          {
            "label": "Criminal",
            "value": "38.6%"
          },
          {
            "label": "Family",
            "value": "31.8%"
          },
          {
            "label": "Health",
            "value": "76.7%"
          },
          {
            "label": "Immigration",
            "value": "27.3%"
          },
          {
            "label": "Weighted Pass",
            "value": "87.7%"
          }
        ]
      },
      {
        "id": "zai-glm-5-3-flash",
        "label": "zai/glm-5.3-flash",
        "model": "zai/glm-5.3-flash",
        "provider": "Zhipu AI",
        "harness": null,
        "effort": "max",
        "score": 45.192,
        "costUsd": 0.120196,
        "uncertainty": {
          "lower": 41.733,
          "upper": 48.651,
          "label": "±1 standard error"
        },
        "sourceUrl": "https://www.vals.ai/benchmarks/legal_research",
        "details": [
          {
            "label": "Publisher model id",
            "value": "zai/glm-5.3-flash"
          },
          {
            "label": "Standard error",
            "value": "±3.46 points"
          },
          {
            "label": "Mean latency",
            "value": "3,067 seconds per test"
          },
          {
            "label": "Evaluation mode",
            "value": "Agentic · many steps allowed"
          },
          {
            "label": "Administrative / Regulatory",
            "value": "57.6%"
          },
          {
            "label": "Business & Commercial",
            "value": "44.2%"
          },
          {
            "label": "Civil Litigation",
            "value": "40.7%"
          },
          {
            "label": "Constitutional / Civil Rights",
            "value": "44.7%"
          },
          {
            "label": "Criminal",
            "value": "33.3%"
          },
          {
            "label": "Family",
            "value": "27.3%"
          },
          {
            "label": "Health",
            "value": "73.3%"
          },
          {
            "label": "Immigration",
            "value": "45.5%"
          },
          {
            "label": "Weighted Pass",
            "value": "87.5%"
          }
        ]
      }
    ]
  }
}
