{
  "name": "The Latent Score",
  "release": "tli-2026-v4.2-2026-09-06",
  "methodologyVersion": "tli-2026-v4.2",
  "dateModified": "2026-09-06",
  "publisher": "The Latent",
  "source": "https://thelatent.co/score/releases/tli-2026-v4.2-2026-09-06",
  "currentRankings": "https://thelatent.co/score",
  "schemaVersion": 2,
  "scale": {
    "mean": 100,
    "sd": 15,
    "unit": "points",
    "description": "Frozen calibration-cohort reference scale; not human IQ or a percentage.",
    "cohortSize": 18
  },
  "definitions": {
    "rank": "Rank by point estimate among models meeting the coverage floor; null means unranked.",
    "stabilityRange": "Published score stability range in points.",
    "domainScore": "Standard-deviation units within a domain relative to the calibration cohort. Null means no measured score; it is not zero.",
    "benchmarksUsed": "Number of results contributing to the score, not a count of independent experiments."
  },
  "domains": [
    {
      "id": "knowledge_reasoning",
      "name": "Knowledge & reasoning",
      "description": "Academic knowledge, mathematics, abstract reasoning, factuality, and long-context reasoning.",
      "unit": "calibration-cohort standard deviations",
      "benchmarkCount": 18
    },
    {
      "id": "coding_swe",
      "name": "Coding & software engineering",
      "description": "Agentic terminal work, issue resolution, long-horizon engineering, competitive coding, and autonomous ML engineering.",
      "unit": "calibration-cohort standard deviations",
      "benchmarkCount": 23
    },
    {
      "id": "agentic_tooluse",
      "name": "Agentic tool & computer use",
      "description": "Tool orchestration, MCP servers, business-workflow automation, browsing, and GUI computer use.",
      "unit": "calibration-cohort standard deviations",
      "benchmarkCount": 21
    },
    {
      "id": "professional_work",
      "name": "Professional & real-world work",
      "description": "Economically valuable knowledge work: legal, finance, tax, medical, enterprise documents, and occupational workflows.",
      "unit": "calibration-cohort standard deviations",
      "benchmarkCount": 22
    },
    {
      "id": "multimodal",
      "name": "Multimodal & vision",
      "description": "Chart, image, video, and handwriting understanding.",
      "unit": "calibration-cohort standard deviations",
      "benchmarkCount": 13
    },
    {
      "id": "security",
      "name": "Cybersecurity",
      "description": "Vulnerability analysis and exploit development capability.",
      "unit": "calibration-cohort standard deviations",
      "benchmarkCount": 3
    },
    {
      "id": "preference_communication",
      "name": "Preference & communication",
      "description": "Arena-style human preference, emotional intelligence, creative writing, and design preference.",
      "unit": "calibration-cohort standard deviations",
      "benchmarkCount": 4
    }
  ],
  "methodology": "https://thelatent.co/methodology",
  "license": "https://thelatent.co/terms-service#data-usage",
  "models": [
    {
      "modelId": "fable-5-1",
      "evidenceUrl": "https://thelatent.co/score/releases/tli-2026-v4.2-2026-09-06/fable-5-1",
      "currentEvidenceUrl": "https://thelatent.co/score/models/fable-5-1",
      "name": "Fable 5.1",
      "provider": "Anthropic",
      "rank": 1,
      "score": 128.2,
      "stabilityRange": [
        122.7,
        134.2
      ],
      "provisional": false,
      "benchmarksUsed": 57,
      "domains": [
        {
          "domainId": "knowledge_reasoning",
          "score": 2,
          "benchmarksObserved": 13
        },
        {
          "domainId": "coding_swe",
          "score": 1.47,
          "benchmarksObserved": 15
        },
        {
          "domainId": "agentic_tooluse",
          "score": 1.77,
          "benchmarksObserved": 8
        },
        {
          "domainId": "professional_work",
          "score": 1.81,
          "benchmarksObserved": 15
        },
        {
          "domainId": "multimodal",
          "score": 0.29,
          "benchmarksObserved": 5
        },
        {
          "domainId": "security",
          "score": null,
          "benchmarksObserved": 0
        },
        {
          "domainId": "preference_communication",
          "score": 1.4,
          "benchmarksObserved": 1
        }
      ]
    },
    {
      "modelId": "gpt-6-astra",
      "evidenceUrl": "https://thelatent.co/score/releases/tli-2026-v4.2-2026-09-06/gpt-6-astra",
      "currentEvidenceUrl": "https://thelatent.co/score/models/gpt-6-astra",
      "name": "GPT-6 Astra",
      "provider": "OpenAI",
      "rank": 2,
      "score": 124.4,
      "stabilityRange": [
        114.1,
        150
      ],
      "provisional": false,
      "benchmarksUsed": 44,
      "domains": [
        {
          "domainId": "knowledge_reasoning",
          "score": 2.17,
          "benchmarksObserved": 11
        },
        {
          "domainId": "coding_swe",
          "score": 1.49,
          "benchmarksObserved": 11
        },
        {
          "domainId": "agentic_tooluse",
          "score": 1.7,
          "benchmarksObserved": 7
        },
        {
          "domainId": "professional_work",
          "score": -0.17,
          "benchmarksObserved": 10
        },
        {
          "domainId": "multimodal",
          "score": 0.12,
          "benchmarksObserved": 4
        },
        {
          "domainId": "security",
          "score": 3.34,
          "benchmarksObserved": 1
        },
        {
          "domainId": "preference_communication",
          "score": null,
          "benchmarksObserved": 0
        }
      ]
    },
    {
      "modelId": "opus-5",
      "evidenceUrl": "https://thelatent.co/score/releases/tli-2026-v4.2-2026-09-06/opus-5",
      "currentEvidenceUrl": "https://thelatent.co/score/models/opus-5",
      "name": "Opus 5",
      "provider": "Anthropic",
      "rank": 3,
      "score": 121.4,
      "stabilityRange": [
        115,
        126.2
      ],
      "provisional": false,
      "benchmarksUsed": 84,
      "domains": [
        {
          "domainId": "knowledge_reasoning",
          "score": 1.23,
          "benchmarksObserved": 17
        },
        {
          "domainId": "coding_swe",
          "score": 1.12,
          "benchmarksObserved": 20
        },
        {
          "domainId": "agentic_tooluse",
          "score": 1.3,
          "benchmarksObserved": 16
        },
        {
          "domainId": "professional_work",
          "score": 1.9,
          "benchmarksObserved": 20
        },
        {
          "domainId": "multimodal",
          "score": 0.3,
          "benchmarksObserved": 6
        },
        {
          "domainId": "security",
          "score": 0.71,
          "benchmarksObserved": 1
        },
        {
          "domainId": "preference_communication",
          "score": 0.93,
          "benchmarksObserved": 4
        }
      ]
    },
    {
      "modelId": "fable-5",
      "evidenceUrl": "https://thelatent.co/score/releases/tli-2026-v4.2-2026-09-06/fable-5",
      "currentEvidenceUrl": "https://thelatent.co/score/models/fable-5",
      "name": "Fable 5",
      "provider": "Anthropic",
      "rank": 4,
      "score": 118.1,
      "stabilityRange": [
        112.9,
        122.6
      ],
      "provisional": false,
      "benchmarksUsed": 96,
      "domains": [
        {
          "domainId": "knowledge_reasoning",
          "score": 1.33,
          "benchmarksObserved": 17
        },
        {
          "domainId": "coding_swe",
          "score": 0.98,
          "benchmarksObserved": 21
        },
        {
          "domainId": "agentic_tooluse",
          "score": 0.82,
          "benchmarksObserved": 18
        },
        {
          "domainId": "professional_work",
          "score": 1.11,
          "benchmarksObserved": 21
        },
        {
          "domainId": "multimodal",
          "score": 1.37,
          "benchmarksObserved": 12
        },
        {
          "domainId": "security",
          "score": 0.91,
          "benchmarksObserved": 3
        },
        {
          "domainId": "preference_communication",
          "score": 1.06,
          "benchmarksObserved": 4
        }
      ]
    },
    {
      "modelId": "muse-spark-1-3",
      "evidenceUrl": "https://thelatent.co/score/releases/tli-2026-v4.2-2026-09-06/muse-spark-1-3",
      "currentEvidenceUrl": "https://thelatent.co/score/models/muse-spark-1-3",
      "name": "Muse Spark 1.3",
      "provider": "Meta",
      "rank": 5,
      "score": 111.9,
      "stabilityRange": [
        105.5,
        118.4
      ],
      "provisional": false,
      "benchmarksUsed": 22,
      "domains": [
        {
          "domainId": "knowledge_reasoning",
          "score": 0.53,
          "benchmarksObserved": 6
        },
        {
          "domainId": "coding_swe",
          "score": 1.22,
          "benchmarksObserved": 4
        },
        {
          "domainId": "agentic_tooluse",
          "score": 0.55,
          "benchmarksObserved": 5
        },
        {
          "domainId": "professional_work",
          "score": 0.79,
          "benchmarksObserved": 4
        },
        {
          "domainId": "multimodal",
          "score": -0.12,
          "benchmarksObserved": 3
        },
        {
          "domainId": "security",
          "score": null,
          "benchmarksObserved": 0
        },
        {
          "domainId": "preference_communication",
          "score": null,
          "benchmarksObserved": 0
        }
      ]
    },
    {
      "modelId": "gpt-5-6-sol",
      "evidenceUrl": "https://thelatent.co/score/releases/tli-2026-v4.2-2026-09-06/gpt-5-6-sol",
      "currentEvidenceUrl": "https://thelatent.co/score/models/gpt-5-6-sol",
      "name": "GPT-5.6 Sol",
      "provider": "OpenAI",
      "rank": 6,
      "score": 108.1,
      "stabilityRange": [
        103.7,
        114.4
      ],
      "provisional": false,
      "benchmarksUsed": 103,
      "domains": [
        {
          "domainId": "knowledge_reasoning",
          "score": 0.73,
          "benchmarksObserved": 18
        },
        {
          "domainId": "coding_swe",
          "score": 0.77,
          "benchmarksObserved": 22
        },
        {
          "domainId": "agentic_tooluse",
          "score": 0.59,
          "benchmarksObserved": 21
        },
        {
          "domainId": "professional_work",
          "score": 0.02,
          "benchmarksObserved": 22
        },
        {
          "domainId": "multimodal",
          "score": 0.98,
          "benchmarksObserved": 13
        },
        {
          "domainId": "security",
          "score": 0.87,
          "benchmarksObserved": 3
        },
        {
          "domainId": "preference_communication",
          "score": 0.17,
          "benchmarksObserved": 4
        }
      ]
    },
    {
      "modelId": "kimi-k3",
      "evidenceUrl": "https://thelatent.co/score/releases/tli-2026-v4.2-2026-09-06/kimi-k3",
      "currentEvidenceUrl": "https://thelatent.co/score/models/kimi-k3",
      "name": "Kimi K3",
      "provider": "Moonshot AI",
      "rank": 7,
      "score": 104.7,
      "stabilityRange": [
        98.6,
        109.3
      ],
      "provisional": false,
      "benchmarksUsed": 87,
      "domains": [
        {
          "domainId": "knowledge_reasoning",
          "score": -0.12,
          "benchmarksObserved": 16
        },
        {
          "domainId": "coding_swe",
          "score": 0.13,
          "benchmarksObserved": 19
        },
        {
          "domainId": "agentic_tooluse",
          "score": 0.44,
          "benchmarksObserved": 16
        },
        {
          "domainId": "professional_work",
          "score": 0.38,
          "benchmarksObserved": 19
        },
        {
          "domainId": "multimodal",
          "score": 0.76,
          "benchmarksObserved": 10
        },
        {
          "domainId": "security",
          "score": -0.11,
          "benchmarksObserved": 3
        },
        {
          "domainId": "preference_communication",
          "score": 0.93,
          "benchmarksObserved": 4
        }
      ]
    },
    {
      "modelId": "glm-5-3",
      "evidenceUrl": "https://thelatent.co/score/releases/tli-2026-v4.2-2026-09-06/glm-5-3",
      "currentEvidenceUrl": "https://thelatent.co/score/models/glm-5-3",
      "name": "GLM-5.3",
      "provider": "Z.ai",
      "rank": 8,
      "score": 103.2,
      "stabilityRange": [
        95.4,
        108.4
      ],
      "provisional": false,
      "benchmarksUsed": 55,
      "domains": [
        {
          "domainId": "knowledge_reasoning",
          "score": -0.57,
          "benchmarksObserved": 11
        },
        {
          "domainId": "coding_swe",
          "score": 0.67,
          "benchmarksObserved": 16
        },
        {
          "domainId": "agentic_tooluse",
          "score": 0.42,
          "benchmarksObserved": 9
        },
        {
          "domainId": "professional_work",
          "score": 0.53,
          "benchmarksObserved": 12
        },
        {
          "domainId": "multimodal",
          "score": -0.2,
          "benchmarksObserved": 2
        },
        {
          "domainId": "security",
          "score": 0.38,
          "benchmarksObserved": 3
        },
        {
          "domainId": "preference_communication",
          "score": 0.27,
          "benchmarksObserved": 2
        }
      ]
    },
    {
      "modelId": "gemini-3-8-flash",
      "evidenceUrl": "https://thelatent.co/score/releases/tli-2026-v4.2-2026-09-06/gemini-3-8-flash",
      "currentEvidenceUrl": "https://thelatent.co/score/models/gemini-3-8-flash",
      "name": "Gemini 3.8 Flash",
      "provider": "Google",
      "rank": 9,
      "score": 102.5,
      "stabilityRange": [
        94,
        110.4
      ],
      "provisional": false,
      "benchmarksUsed": 50,
      "domains": [
        {
          "domainId": "knowledge_reasoning",
          "score": 0.25,
          "benchmarksObserved": 10
        },
        {
          "domainId": "coding_swe",
          "score": 0.33,
          "benchmarksObserved": 14
        },
        {
          "domainId": "agentic_tooluse",
          "score": -0.17,
          "benchmarksObserved": 4
        },
        {
          "domainId": "professional_work",
          "score": -0.4,
          "benchmarksObserved": 14
        },
        {
          "domainId": "multimodal",
          "score": -1.67,
          "benchmarksObserved": 7
        },
        {
          "domainId": "security",
          "score": null,
          "benchmarksObserved": 0
        },
        {
          "domainId": "preference_communication",
          "score": 0.91,
          "benchmarksObserved": 1
        }
      ]
    },
    {
      "modelId": "muse-spark-1-2",
      "evidenceUrl": "https://thelatent.co/score/releases/tli-2026-v4.2-2026-09-06/muse-spark-1-2",
      "currentEvidenceUrl": "https://thelatent.co/score/models/muse-spark-1-2",
      "name": "Muse Spark 1.2",
      "provider": "Meta",
      "rank": 10,
      "score": 102.2,
      "stabilityRange": [
        93.1,
        107.3
      ],
      "provisional": false,
      "benchmarksUsed": 49,
      "domains": [
        {
          "domainId": "knowledge_reasoning",
          "score": -0.12,
          "benchmarksObserved": 9
        },
        {
          "domainId": "coding_swe",
          "score": -0.47,
          "benchmarksObserved": 8
        },
        {
          "domainId": "agentic_tooluse",
          "score": 0.33,
          "benchmarksObserved": 10
        },
        {
          "domainId": "professional_work",
          "score": 0.39,
          "benchmarksObserved": 15
        },
        {
          "domainId": "multimodal",
          "score": 0.11,
          "benchmarksObserved": 3
        },
        {
          "domainId": "security",
          "score": -0.19,
          "benchmarksObserved": 1
        },
        {
          "domainId": "preference_communication",
          "score": 0.7,
          "benchmarksObserved": 3
        }
      ]
    },
    {
      "modelId": "grok-4-6-high",
      "evidenceUrl": "https://thelatent.co/score/releases/tli-2026-v4.2-2026-09-06/grok-4-6-high",
      "currentEvidenceUrl": "https://thelatent.co/score/models/grok-4-6-high",
      "name": "Grok 4.6",
      "provider": "xAI",
      "rank": 11,
      "score": 102,
      "stabilityRange": [
        97,
        108.5
      ],
      "provisional": false,
      "benchmarksUsed": 57,
      "domains": [
        {
          "domainId": "knowledge_reasoning",
          "score": -0.09,
          "benchmarksObserved": 13
        },
        {
          "domainId": "coding_swe",
          "score": 0.45,
          "benchmarksObserved": 15
        },
        {
          "domainId": "agentic_tooluse",
          "score": 1.04,
          "benchmarksObserved": 6
        },
        {
          "domainId": "professional_work",
          "score": 0.33,
          "benchmarksObserved": 16
        },
        {
          "domainId": "multimodal",
          "score": -2.38,
          "benchmarksObserved": 4
        },
        {
          "domainId": "security",
          "score": 0.07,
          "benchmarksObserved": 2
        },
        {
          "domainId": "preference_communication",
          "score": -0.58,
          "benchmarksObserved": 1
        }
      ]
    },
    {
      "modelId": "qwen-3-8-max",
      "evidenceUrl": "https://thelatent.co/score/releases/tli-2026-v4.2-2026-09-06/qwen-3-8-max",
      "currentEvidenceUrl": "https://thelatent.co/score/models/qwen-3-8-max",
      "name": "Qwen3.8-Max",
      "provider": "Alibaba",
      "rank": 12,
      "score": 97.1,
      "stabilityRange": [
        89.7,
        101.5
      ],
      "provisional": false,
      "benchmarksUsed": 71,
      "domains": [
        {
          "domainId": "knowledge_reasoning",
          "score": -0.23,
          "benchmarksObserved": 11
        },
        {
          "domainId": "coding_swe",
          "score": -0.9,
          "benchmarksObserved": 14
        },
        {
          "domainId": "agentic_tooluse",
          "score": -0.39,
          "benchmarksObserved": 11
        },
        {
          "domainId": "professional_work",
          "score": 0.05,
          "benchmarksObserved": 16
        },
        {
          "domainId": "multimodal",
          "score": 2.17,
          "benchmarksObserved": 13
        },
        {
          "domainId": "security",
          "score": -0.18,
          "benchmarksObserved": 3
        },
        {
          "domainId": "preference_communication",
          "score": 0.54,
          "benchmarksObserved": 3
        }
      ]
    },
    {
      "modelId": "gemini-3-7-flash",
      "evidenceUrl": "https://thelatent.co/score/releases/tli-2026-v4.2-2026-09-06/gemini-3-7-flash",
      "currentEvidenceUrl": "https://thelatent.co/score/models/gemini-3-7-flash",
      "name": "Gemini 3.7 Flash",
      "provider": "Google",
      "rank": 13,
      "score": 95,
      "stabilityRange": [
        86.4,
        105
      ],
      "provisional": false,
      "benchmarksUsed": 60,
      "domains": [
        {
          "domainId": "knowledge_reasoning",
          "score": 0.14,
          "benchmarksObserved": 15
        },
        {
          "domainId": "coding_swe",
          "score": -0.26,
          "benchmarksObserved": 15
        },
        {
          "domainId": "agentic_tooluse",
          "score": -1.29,
          "benchmarksObserved": 5
        },
        {
          "domainId": "professional_work",
          "score": -0.9,
          "benchmarksObserved": 15
        },
        {
          "domainId": "multimodal",
          "score": -0.05,
          "benchmarksObserved": 7
        },
        {
          "domainId": "security",
          "score": 0.75,
          "benchmarksObserved": 1
        },
        {
          "domainId": "preference_communication",
          "score": 0.34,
          "benchmarksObserved": 2
        }
      ]
    },
    {
      "modelId": "gpt-5-6-terra",
      "evidenceUrl": "https://thelatent.co/score/releases/tli-2026-v4.2-2026-09-06/gpt-5-6-terra",
      "currentEvidenceUrl": "https://thelatent.co/score/models/gpt-5-6-terra",
      "name": "GPT-5.6 Terra",
      "provider": "OpenAI",
      "rank": 14,
      "score": 94.5,
      "stabilityRange": [
        90.2,
        102.6
      ],
      "provisional": false,
      "benchmarksUsed": 74,
      "domains": [
        {
          "domainId": "knowledge_reasoning",
          "score": -0.28,
          "benchmarksObserved": 16
        },
        {
          "domainId": "coding_swe",
          "score": 0.13,
          "benchmarksObserved": 17
        },
        {
          "domainId": "agentic_tooluse",
          "score": -0.27,
          "benchmarksObserved": 12
        },
        {
          "domainId": "professional_work",
          "score": -0.64,
          "benchmarksObserved": 17
        },
        {
          "domainId": "multimodal",
          "score": -0.4,
          "benchmarksObserved": 6
        },
        {
          "domainId": "security",
          "score": 0.33,
          "benchmarksObserved": 2
        },
        {
          "domainId": "preference_communication",
          "score": -0.66,
          "benchmarksObserved": 4
        }
      ]
    },
    {
      "modelId": "deepseek-v4-pro-max",
      "evidenceUrl": "https://thelatent.co/score/releases/tli-2026-v4.2-2026-09-06/deepseek-v4-pro-max",
      "currentEvidenceUrl": "https://thelatent.co/score/models/deepseek-v4-pro-max",
      "name": "DeepSeek V4 Pro",
      "provider": "DeepSeek",
      "rank": 15,
      "score": 89.4,
      "stabilityRange": [
        82.6,
        97.5
      ],
      "provisional": false,
      "benchmarksUsed": 53,
      "domains": [
        {
          "domainId": "knowledge_reasoning",
          "score": -0.61,
          "benchmarksObserved": 15
        },
        {
          "domainId": "coding_swe",
          "score": -1.02,
          "benchmarksObserved": 13
        },
        {
          "domainId": "agentic_tooluse",
          "score": -0.21,
          "benchmarksObserved": 9
        },
        {
          "domainId": "professional_work",
          "score": -1,
          "benchmarksObserved": 14
        },
        {
          "domainId": "multimodal",
          "score": null,
          "benchmarksObserved": 0
        },
        {
          "domainId": "security",
          "score": 0.59,
          "benchmarksObserved": 1
        },
        {
          "domainId": "preference_communication",
          "score": -0.53,
          "benchmarksObserved": 1
        }
      ]
    },
    {
      "modelId": "sonnet-5",
      "evidenceUrl": "https://thelatent.co/score/releases/tli-2026-v4.2-2026-09-06/sonnet-5",
      "currentEvidenceUrl": "https://thelatent.co/score/models/sonnet-5",
      "name": "Sonnet 5",
      "provider": "Anthropic",
      "rank": 16,
      "score": 88.5,
      "stabilityRange": [
        82.1,
        93.9
      ],
      "provisional": false,
      "benchmarksUsed": 67,
      "domains": [
        {
          "domainId": "knowledge_reasoning",
          "score": -0.49,
          "benchmarksObserved": 11
        },
        {
          "domainId": "coding_swe",
          "score": -0.27,
          "benchmarksObserved": 16
        },
        {
          "domainId": "agentic_tooluse",
          "score": -0.99,
          "benchmarksObserved": 11
        },
        {
          "domainId": "professional_work",
          "score": -0.45,
          "benchmarksObserved": 18
        },
        {
          "domainId": "multimodal",
          "score": -0.51,
          "benchmarksObserved": 6
        },
        {
          "domainId": "security",
          "score": -3.01,
          "benchmarksObserved": 1
        },
        {
          "domainId": "preference_communication",
          "score": -0.68,
          "benchmarksObserved": 4
        }
      ]
    },
    {
      "modelId": "gpt-5-6-luna",
      "evidenceUrl": "https://thelatent.co/score/releases/tli-2026-v4.2-2026-09-06/gpt-5-6-luna",
      "currentEvidenceUrl": "https://thelatent.co/score/models/gpt-5-6-luna",
      "name": "GPT-5.6 Luna",
      "provider": "OpenAI",
      "rank": 17,
      "score": 86,
      "stabilityRange": [
        81.6,
        94.5
      ],
      "provisional": false,
      "benchmarksUsed": 71,
      "domains": [
        {
          "domainId": "knowledge_reasoning",
          "score": -0.94,
          "benchmarksObserved": 15
        },
        {
          "domainId": "coding_swe",
          "score": -0.19,
          "benchmarksObserved": 14
        },
        {
          "domainId": "agentic_tooluse",
          "score": -1.02,
          "benchmarksObserved": 12
        },
        {
          "domainId": "professional_work",
          "score": -0.91,
          "benchmarksObserved": 17
        },
        {
          "domainId": "multimodal",
          "score": -0.44,
          "benchmarksObserved": 7
        },
        {
          "domainId": "security",
          "score": -0.09,
          "benchmarksObserved": 2
        },
        {
          "domainId": "preference_communication",
          "score": -1.21,
          "benchmarksObserved": 4
        }
      ]
    },
    {
      "modelId": "deepseek-v4-flash",
      "evidenceUrl": "https://thelatent.co/score/releases/tli-2026-v4.2-2026-09-06/deepseek-v4-flash",
      "currentEvidenceUrl": "https://thelatent.co/score/models/deepseek-v4-flash",
      "name": "DeepSeek V4 Flash",
      "provider": "DeepSeek",
      "rank": 18,
      "score": 72.9,
      "stabilityRange": [
        67.7,
        82.8
      ],
      "provisional": false,
      "benchmarksUsed": 53,
      "domains": [
        {
          "domainId": "knowledge_reasoning",
          "score": -1.2,
          "benchmarksObserved": 15
        },
        {
          "domainId": "coding_swe",
          "score": -1.5,
          "benchmarksObserved": 12
        },
        {
          "domainId": "agentic_tooluse",
          "score": -2.22,
          "benchmarksObserved": 9
        },
        {
          "domainId": "professional_work",
          "score": -1.62,
          "benchmarksObserved": 11
        },
        {
          "domainId": "multimodal",
          "score": -0.39,
          "benchmarksObserved": 2
        },
        {
          "domainId": "security",
          "score": -0.38,
          "benchmarksObserved": 2
        },
        {
          "domainId": "preference_communication",
          "score": -2.14,
          "benchmarksObserved": 2
        }
      ]
    },
    {
      "modelId": "qwen-3-8-27b",
      "evidenceUrl": "https://thelatent.co/score/releases/tli-2026-v4.2-2026-09-06/qwen-3-8-27b",
      "currentEvidenceUrl": "https://thelatent.co/score/models/qwen-3-8-27b",
      "name": "Qwen3.8-27B",
      "provider": "Alibaba",
      "rank": 19,
      "score": 69.5,
      "stabilityRange": [
        59.4,
        79.9
      ],
      "provisional": false,
      "benchmarksUsed": 51,
      "domains": [
        {
          "domainId": "knowledge_reasoning",
          "score": -2.14,
          "benchmarksObserved": 10
        },
        {
          "domainId": "coding_swe",
          "score": -2.76,
          "benchmarksObserved": 12
        },
        {
          "domainId": "agentic_tooluse",
          "score": -1.09,
          "benchmarksObserved": 7
        },
        {
          "domainId": "professional_work",
          "score": -1.26,
          "benchmarksObserved": 13
        },
        {
          "domainId": "multimodal",
          "score": 0.13,
          "benchmarksObserved": 7
        },
        {
          "domainId": "security",
          "score": null,
          "benchmarksObserved": 0
        },
        {
          "domainId": "preference_communication",
          "score": -1.93,
          "benchmarksObserved": 2
        }
      ]
    }
  ]
}