{
  "level1": [
    {
      "agent": "DeepSeek Agent",
      "model": "DeepSeek-V4-Pro",
      "model_release_date": "2026-08-13",
      "include_in_plot": true,
      "score_10": 0.833,
      "date": "2026-08-13",
      "source": "DeepSeek",
      "source_url": "https://huggingface.co/deepseek-ai/DeepSeek-V4-Pro-0813",
      "trials": 1,
      "focus": "model"
    },
    {
      "agent": "Grok Build",
      "model": "Grok 4.5",
      "model_release_date": "2026-07-14",
      "include_in_plot": true,
      "score_10": 0.79,
      "date": "2026-07-14",
      "source": "xAI",
      "source_url": "https://cursor.com/resources/grok-4-5-model-card.pdf",
      "trials": 1,
      "focus": "model"
    },
    {
      "agent": "Grok Build",
      "model": "Grok 4.6",
      "model_release_date": "2026-08-12",
      "include_in_plot": true,
      "score_10": 0.797,
      "date": "2026-08-12",
      "source": "xAI",
      "source_url": "https://media.x.ai/v1/website/card-4p6-4cd2dc57.pdf",
      "trials": 1,
      "focus": "model"
    },
    {
      "agent": "Claude Code",
      "model": "GLM-5.2",
      "model_release_date": "2026-07-21",
      "include_in_plot": true,
      "score_10": 0.772,
      "date": "2026-07-21",
      "source": "Zhipu AI",
      "source_url": "https://z.ai/blog/glm-5.3",
      "trials": 1,
      "focus": "model"
    },
    {
      "agent": "Claude Code",
      "model": "GLM-5.3",
      "model_release_date": "2026-08-14",
      "include_in_plot": true,
      "score_10": 0.845,
      "date": "2026-08-14",
      "source": "Zhipu AI",
      "source_url": "https://z.ai/blog/glm-5.3",
      "trials": 1,
      "focus": "model"
    },
    {
      "agent": "Claude Code",
      "model": "Feyospace-v1.1 (finetuned from Qwen3.8-27B)",
      "model_release_date": "2026-09-23",
      "include_in_plot": true,
      "icon": "/assets/images/cybergym/feyospace.png",
      "score_10": 0.8441,
      "date": "2026-09-23",
      "source": "Vera Praxis Lab",
      "source_url": "https://verapraxis.ai/en/research/feyospace-v1-1",
      "trials": 1,
      "focus": "model",
      "subRows": [
        {
          "model": "Feyospace-v1",
          "score_10": 0.6304,
          "date": "2026-09-08",
          "source_url": "https://arxiv.org/abs/2609.08418"
        },
        {
          "model": "Feyospace-v0",
          "score_10": 0.5116,
          "date": "2026-09-08",
          "source_url": "https://arxiv.org/abs/2609.08418"
        }
      ]
    },
    {
      "agent": "Claude Code",
      "model": "ASL-Cyber-Flash (finetuned from DeepSeek-V4-Flash)",
      "model_release_date": "2026-09-03",
      "include_in_plot": true,
      "icon": "/assets/images/cybergym/aeye.png",
      "features": [
        "dynamic (uses cleaned vulnerable Docker image)"
      ],
      "score_10": 0.8587,
      "date": "2026-09-03",
      "source": "Aeye Security Lab",
      "source_url": "https://github.com/aeyesec/asl-cyber-flash-submission/",
      "trials": 1,
      "focus": "model"
    },
    {
      "agent": "Google Antigravity",
      "model": "Gemini 3.8 Flash Cyber",
      "model_release_date": "2026-09-02",
      "include_in_plot": true,
      "score_10": 0.8626,
      "date": "2026-09-10",
      "source": "Google",
      "source_url": "https://blog.google/innovation-and-ai/models-and-research/gemini-models/3-8-flash-and-3-8-flash-cyber/",
      "trials": 1,
      "focus": "model"
    },
    {
      "agent": "Panshi \\(磐石\\)",
      "model": "DeepSeek-V4-Pro",
      "model_release_date": "2026-08-13",
      "include_in_plot": true,
      "plot_links": [
        "DeepSeek-V4-Pro"
      ],
      "icon": "/assets/images/cybergym/panshi.png",
      "features": [
        "dynamic (uses cleaned vulnerable Docker image)"
      ],
      "score_10": 0.9542,
      "date": "2026-09-04",
      "source": "Huawei Cloud",
      "source_url": "https://github.com/PanshiAgent/Panshi-Security-Agent",
      "trials": 1,
      "focus": "agent"
    },
    {
      "agent": "NASH",
      "model": "DeepSeek-V4-Flash",
      "model_release_date": "2026-07-31",
      "include_in_plot": true,
      "plot_links": [
        "DeepSeek-V4-Flash"
      ],
      "icon": "/assets/images/cybergym/nash.png",
      "features": [
        "dynamic (uses cleaned vulnerable Docker image)"
      ],
      "score_10": 0.9137,
      "date": "2026-09-01",
      "source": "NASH Team",
      "source_url": "https://github.com/NASH-nrxt/NASH",
      "trials": 1,
      "focus": "agent"
    },
    {
      "agent": "Lyrie Agent",
      "model": "Multi-model (DeepSeek-V4.1-Flash, Abliterated GLM-5.2)",
      "model_release_date": "2026-09-15",
      "include_in_plot": true,
      "plot_links": [
        "GLM-5.2"
      ],
      "icon": "/assets/images/cybergym/lyrie.png",
      "features": [
        "Multi-model",
        "dynamic (uses cleaned vulnerable Docker image)"
      ],
      "score_10": 0.992,
      "date": "2026-09-15",
      "source": "Lyrie",
      "source_url": "https://lyrie.ai/cybergym",
      "trials": 1,
      "focus": "agent"
    },
    {
      "agent": "PwnBot",
      "model": "DeepSeek-V4-Flash",
      "model_release_date": "2026-07-31",
      "include_in_plot": true,
      "plot_links": [
        "DeepSeek-V4-Flash"
      ],
      "icon": "/assets/images/cybergym/pwnbot.png",
      "features": [
        "dynamic (uses cleaned vulnerable Docker image)",
        "test-time mem. (with a test-time-updated knowledge base)"
      ],
      "score_10": 0.931,
      "date": "2026-09-27",
      "source": "FUNNY_04",
      "source_url": "https://github.com/CyberLadX64/PwnBot",
      "trials": 1,
      "focus": "agent"
    },
    {
      "agent": "OrcaCyber",
      "model": "Multi-model (DeepSeek-V4-Flash, OrcaCyber-Zero-1.0)",
      "model_release_date": "2026-07-31",
      "include_in_plot": true,
      "plot_links": [
        "DeepSeek-V4-Flash"
      ],
      "icon": "/assets/images/cybergym/orcacyber.png",
      "features": [
        "Multi-model",
        "dynamic (uses cleaned vulnerable Docker image)"
      ],
      "score_10": 0.9801,
      "date": "2026-09-14",
      "source": "Continuum AI",
      "source_url": "https://github.com/Continuum-AI-Corp/OrcaCyber-Harness-CyberGym-Technical-Report",
      "trials": 1,
      "focus": "agent"
    },
    {
      "agent": "RedbudAI",
      "model": "GLM-5.2",
      "model_release_date": "2026-08-10",
      "include_in_plot": true,
      "icon": "/assets/images/cybergym/redbud.png",
      "features": [
        "dynamic (uses cleaned vulnerable Docker image)",
        "test-time mem. (with a test-time-updated knowledge base)"
      ],
      "score_10": 0.86264,
      "date": "2026-08-10",
      "source": "Redbud",
      "source_url": "https://redbud.tsinghua.edu.cn/posts/cybergym.html",
      "trials": 1,
      "focus": "agent"
    },
    {
      "agent": "DoGNAVY",
      "model": "GLM-5.3",
      "model_release_date": "2026-08-14",
      "include_in_plot": true,
      "icon": "/assets/images/cybergym/dognavy.png",
      "features": [
        "Multi-agent",
        "Memory",
        "dynamic (uses cleaned vulnerable Docker image)"
      ],
      "score_10": 0.9496,
      "date": "2026-08-17",
      "source": "deepsec@DARKNAVY",
      "source_url": "https://deepsec.darknavy.net/blog/cybergym",
      "trials": 1,
      "focus": "agent",
      "subRows": [
        { "model": "GLM-5.2", "score_10": 0.9084, "date": "2026-08-03" }
      ]
    },
    {
      "agent": "NSFOCUS AI",
      "model": "GLM-5.3",
      "model_release_date": "2026-08-14",
      "include_in_plot": true,
      "icon": "/assets/images/cybergym/nsfocus.png",
      "features": ["dynamic (uses cleaned vulnerable Docker image)"],
      "score_10": 0.9502,
      "date": "2026-08-13",
      "source": "NSFOCUS",
      "source_url": "https://nsfocusglobal.com/nsfocus-ai-cybergym-level-1-technical-report/",
      "trials": 1,
      "focus": "agent"
    },
    {
      "agent": "Gcsa Agent",
      "model": "Multi-model (Grok 4.5, Grok 4.6)",
      "model_release_date": "2026-08-12",
      "include_in_plot": true,
      "plot_links": [
        "Grok 4.5",
        "Grok 4.6"
      ],
      "icon": "/assets/images/cybergym/gcsa.png",
      "features": [
        "Multi-model",
        "dynamic (uses cleaned vulnerable Docker image)"
      ],
      "score_10": 0.9131,
      "date": "2026-08-25",
      "source": "Gcsa",
      "source_url": "https://gcsa.org/cybergym/l1-writeup-v2.0.html",
      "trials": 1,
      "focus": "agent"
    },
    {
      "agent": "mneme",
      "model": "Claude Opus 4.6",
      "model_release_date": "2026-02-05",
      "include_in_plot": true,
      "plot_links": [
        "Claude Opus 4.6"
      ],
      "icon": "/assets/images/cybergym/mneme.png",
      "features": [
        "dynamic (uses cleaned vulnerable Docker image)"
      ],
      "score_10": 0.9104,
      "date": "2026-07-13",
      "source": "Soongsil AISC",
      "source_url": "https://aisc-ssu.github.io/mneme/",
      "trials": 1,
      "focus": "agent"
    },
    {
      "agent": "VARAS-OneMind \\(智安·宇盾\\)",
      "model": "DeepSeek-V4-Flash",
      "model_release_date": "2026-07-31",
      "include_in_plot": true,
      "plot_links": [
        "DeepSeek-V4-Flash"
      ],
      "icon": "/assets/images/cybergym/varas.png",
      "features": [
        "dynamic (uses cleaned vulnerable Docker image)",
        "test-time mem. (with a test-time-updated knowledge base)"
      ],
      "score_10": 0.9695,
      "date": "2026-09-16",
      "source": "VARAS",
      "source_url": "https://varas-onemind.github.io/posts/cybergym_v2/",
      "trials": 1,
      "focus": "agent",
      "subRows": [
        {
          "model": "GPT-5.4",
          "score_10": 0.7379,
          "date": "2026-08-14",
          "source_url": "https://varas-onemind.github.io/posts/cybergym/",
          "features": [
            "test-time mem. (with a test-time-updated knowledge base)"
          ]
        }
      ]
    },
    {
      "agent": "Gusion",
      "model": "DeepSeek-V4-Flash",
      "model_release_date": "2026-07-31",
      "include_in_plot": true,
      "plot_links": [
        "DeepSeek-V4-Flash"
      ],
      "icon": "/assets/images/cybergym/gusion.png",
      "features": [],
      "score_10": 0.8706,
      "date": "2026-08-16",
      "source": "Gusion",
      "source_url": "https://github.com/EBWi11/Gusion/blob/add-level1-technical-report/README.md",
      "trials": 1,
      "focus": "agent"
    },
    {
      "agent": "FangcunCyber",
      "model": "DeepSeek-V4-Flash",
      "model_release_date": "2026-07-31",
      "include_in_plot": true,
      "plot_links": [
        "DeepSeek-V4-Flash"
      ],
      "icon": "/assets/images/cybergym/fangcun.png",
      "features": [
        "test-time mem. (with a test-time-updated knowledge base)"
      ],
      "score_10": 0.9131,
      "date": "2026-08-16",
      "source": "Fangcun AI",
      "source_url": "https://github.com/Fangcun-AI/FangcunCyber",
      "trials": 1,
      "focus": "agent"
    },
    {
      "agent": "Alipay AI4SDL",
      "model": "GLM-5.3",
      "model_release_date": "2026-08-14",
      "include_in_plot": true,
      "icon": "/assets/images/cybergym/alipay-logo.png",
      "features": [
        "dynamic (uses cleaned vulnerable Docker image)"
      ],
      "score_10": 0.9675,
      "date": "2026-09-06",
      "source": "Alipay",
      "source_url": "https://github.com/Alipay-Risk-Tech/CyberGym-Writeup",
      "trials": 1,
      "focus": "agent"
    },
    {
      "agent": "Creation \\(天工\\)",
      "model": "Multi-model (Creation Model, DeepSeek-V4-Pro, Qwen 3.8 Max)",
      "model_release_date": "2026-08-30",
      "include_in_plot": true,
      "plot_links": [
        "DeepSeek-V4-Pro"
      ],
      "icon": "/assets/images/cybergym/creation.png",
      "features": [
        "Multi-model",
        "dynamic (uses cleaned vulnerable Docker image)"
      ],
      "score_10": 0.9847,
      "date": "2026-09-07",
      "source": "Creation Lab",
      "source_url": "https://github.com/AntAISecurityLab/Creation",
      "trials": 1,
      "focus": "agent",
      "subRows": [
        {
          "model": "Multi-model (DeepSeek-V4-Pro, Creation Model, Qwen 3.8 Max)",
          "score_10": 0.9535,
          "date": "2026-08-30"
        }
      ]
    },
    {
      "agent": "Spur",
      "model": "Qwen-Internal",
      "model_release_date": "2026-08-29",
      "include_in_plot": true,
      "icon": "/assets/images/cybergym/spur.png",
      "features": [],
      "score_10": 0.918381,
      "date": "2026-09-03",
      "source": "Alibaba Cloud Security",
      "source_url": null,
      "trials": 1,
      "focus": "agent"
    },
    {
      "agent": "Hero Agent",
      "model": "Hero (finetuned from GLM-5.2)",
      "model_release_date": "2026-09-03",
      "include_in_plot": true,
      "plot_links": [
        "GLM-5.2"
      ],
      "icon": "/assets/images/cybergym/dream.png",
      "features": [
        "dynamic (uses cleaned vulnerable Docker image)",
        "test-time mem. (with a test-time-updated knowledge base)"
      ],
      "score_10": 0.9662,
      "date": "2026-09-03",
      "source": "Dream",
      "source_url": "https://www.dreamgroup.com/blog/hercules-on-premises-ai-vulnerability-model-ranked-1-on-cybergym",
      "visible_after": "2026-09-03T07:00:00-07:00",
      "trials": 1,
      "focus": "agent"
    },
    {
      "agent": "RO0T Agent",
      "model": "DeepSeek-V4.1-Flash",
      "model_release_date": "2026-09-18",
      "include_in_plot": true,
      "icon": "/assets/images/cybergym/root-lab.png",
      "features": [
        "dynamic (uses cleaned vulnerable Docker image)"
      ],
      "score_10": 0.977,
      "date": "2026-09-18",
      "source": "RO0T Lab",
      "source_url": "https://github.com/RO0TLab/RO0T-Agent/blob/main/README.md",
      "trials": 1,
      "focus": "agent",
      "subRows": [
        {
          "model": "DeepSeek-V4-Flash",
          "score_10": 0.9416,
          "date": "2026-08-26"
        }
      ]
    },
    {
      "agent": "DeepSeek Agent",
      "model": "DeepSeek-V4-Flash",
      "model_release_date": "2026-07-31",
      "include_in_plot": true,
      "score_10": 0.767,
      "date": "2026-07-31",
      "source": "DeepSeek",
      "source_url": "https://huggingface.co/deepseek-ai/DeepSeek-V4-Flash-0731",
      "trials": 1,
      "focus": "model"
    },
    {
      "agent": "Velldepth Agent",
      "model": "XekRung",
      "model_release_date": "2026-07-23",
      "include_in_plot": true,
      "icon": "/assets/images/cybergym/velldepth.png",
      "score_10": 0.8534,
      "date": "2026-08-03",
      "source": "Alibaba Security",
      "source_url": "https://alibaba-velldepth.github.io/writeups/",
      "trials": 1,
      "focus": "agent"
    },
    {
      "agent": "XekRung Agent",
      "model": "XekRung-1.5-27B-Preview (finetuned from Qwen3.8-27B)",
      "model_release_date": "2026-09-13",
      "include_in_plot": true,
      "icon": "/assets/images/cybergym/velldepth.png",
      "score_10": 0.8892,
      "date": "2026-09-13",
      "source": "Alibaba Security",
      "source_url": "https://alibaba-velldepth.github.io/writeups/model-track.html",
      "trials": 1,
      "focus": "model"
    },
    {
      "agent": "JiuXuan \\(九玄\\)",
      "model": "GLM-5.1",
      "model_release_date": "2026-04-07",
      "include_in_plot": true,
      "plot_links": [
        "GLM-5.1"
      ],
      "icon": "/assets/images/cybergym/jiuxuan.png",
      "features": [
        "Memory",
        "dynamic (uses cleaned vulnerable Docker image)"
      ],
      "score_10": 0.7286,
      "date": "2026-08-03",
      "source": "China Mobile Jiutian AI",
      "source_url": "https://github.com/as837430732/JiuXuan",
      "trials": 1,
      "focus": "agent"
    },
    {
      "agent": "Wiz Atlas",
      "model": "Multi-model (GPT-5.5, Claude Opus 4.6)",
      "model_release_date": "2026-07-27",
      "include_in_plot": true,
      "plot_links": [
        "GPT-5.5",
        "Claude Opus 4.6"
      ],
      "icon": "/assets/images/cybergym/wiz-atlas.png",
      "features": [
        "Multi-model",
        "Multi-stage"
      ],
      "score_10": 0.909,
      "date": "2026-07-27",
      "source": "Wiz",
      "source_url": "https://www.wiz.io/blog/atlas-ai-vulnerability-researcher",
      "trials": 1,
      "focus": "agent"
    },
    {
      "agent": "Sangfor AI",
      "model": "GLM-5.3",
      "model_release_date": "2026-08-14",
      "include_in_plot": true,
      "icon": "/assets/images/cybergym/sangfor.png",
      "features": [
        "Orchestration",
        "Multi-stage",
        "dynamic (uses cleaned vulnerable Docker image)"
      ],
      "score_10": 0.9721,
      "date": "2026-09-02",
      "source": "Sangfor AI",
      "source_url": "https://github.com/Sangfor-AI/cybergym-submission-sangfor-ai-v2",
      "trials": 1,
      "focus": "agent",
      "subRows": [
        {
          "model": "DeepSeek-V4-Flash",
          "score_10": 0.9317,
          "date": "2026-08-08",
          "source_url": "https://github.com/Sangfor-AI/cybergym-submission-sangfor-ai-v2"
        },
        {
          "model": "GLM-5.2",
          "score_10": 0.8633,
          "date": "2026-07-21",
          "source_url": "https://www.sangfor.com/news-and-press-release/sangfor-ai-ranked-4-on-cybergym-2026"
        }
      ]
    },
    {
      "agent": "Whitzard \\(白泽\\)",
      "model": "DeepSeek-V4-Flash",
      "model_release_date": "2026-07-31",
      "include_in_plot": true,
      "icon": "/assets/images/cybergym/fudan-logo.png",
      "features": [
        "dynamic (uses cleaned vulnerable Docker image)"
      ],
      "score_10": 0.912,
      "date": "2026-08-07",
      "source": "Fudan Whitzard",
      "source_url": "https://github.com/WhitzardAgent/Whitzard",
      "trials": 1,
      "focus": "agent",
      "subRows": [
        {
          "model": "GLM-5.1-FP8",
          "score_10": 0.689,
          "date": "2026-07-21"
        }
      ]
    },
    {
      "agent": "Claude Code",
      "model": "DeepSeek-V4-Pro Preview",
      "model_release_date": "2026-04-24",
      "include_in_plot": true,
      "score_10": 0.577,
      "date": "2026-07-13",
      "source": "XDxAI",
      "source_url": "https://github.com/XDxAI/cybergym-deepseek-submission-2026",
      "trials": 1,
      "focus": "model"
    },
    {
      "agent": "XD Agent",
      "model": "DeepSeek-V4-Pro Preview",
      "model_release_date": "2026-04-24",
      "include_in_plot": true,
      "icon": "/assets/images/cybergym/xdagent.png",
      "features": [],
      "score_10": 0.641,
      "date": "2026-08-11",
      "source": "XDxAI",
      "source_url": "https://github.com/XDxAI/XD-Agent/blob/main/CyberGym_submission_XDagent.md",
      "trials": 1,
      "focus": "agent"
    },
    {
      "agent": "Meta Agent",
      "model": "Muse Spark 1.1 (helpful-only)",
      "model_release_date": "2026-07-09",
      "include_in_plot": true,
      "score_10": 0.59,
      "date": "2026-07-09",
      "source": "Meta AI",
      "source_url": "https://ai.meta.com/static-resource/muse-spark-1-1-evaluation-report",
      "trials": 1,
      "focus": "model"
    },
    {
      "agent": "Xuanwu Atuin AI",
      "model": "GLM-5.2",
      "model_release_date": "2026-07-22",
      "include_in_plot": true,
      "score_10": 0.848,
      "date": "2026-07-22",
      "source": "Tencent Xuanwu Lab",
      "source_url": "https://xlab.tencent.com/en/2026/07/17/xuanwu-atuin-cybergym-glm52/",
      "trials": 1,
      "focus": "agent",
      "features": [
        "dynamic (uses cleaned vulnerable Docker image)"
      ],
      "icon": "/assets/images/cybergym/XuanwuLogo.png",
      "subRows": [
        {
          "model": "GLM-5.1",
          "score_10": 0.84,
          "date": "2026-07-02",
          "source_url": "https://xlab.tencent.com/en/2026/07/02/xuanwu-atuin-cybergym/"
        }
      ]
    },
    {
      "agent": "MopMonk Agent",
      "model": "MiniMax M3",
      "model_release_date": "2026-06-01",
      "include_in_plot": true,
      "icon": "/assets/images/cybergym/mopmonk.png",
      "score_10": 0.731,
      "date": "2026-06-28",
      "source": "MopMonk AI",
      "source_url": "https://github.com/MopMonkAI/MopMonkAgent",
      "trials": 1,
      "focus": "agent"
    },
    {
      "agent": "OpenAI Agent",
      "model": "GPT-5.5-Cyber",
      "model_release_date": "2026-06-22",
      "include_in_plot": true,
      "score_10": 0.856,
      "date": "2026-06-22",
      "source": "OpenAI",
      "source_url": "https://openai.com/index/daybreak-securing-the-world/",
      "trials": 1,
      "focus": "model"
    },
    {
      "agent": "Crystalline",
      "model": "Claude Opus 4.6",
      "model_release_date": "2026-02-05",
      "include_in_plot": true,
      "icon": "/assets/images/cybergym/crystalline.png",
      "features": [
        "test-time mem. (with a test-time-updated knowledge base)",
        "Test-time memory"
      ],
      "score_10": 0.896,
      "date": "2026-06-08",
      "source": "Independent researcher",
      "source_url": "https://github.com/synchopate/cybergym-logos",
      "trials": 1,
      "focus": "agent"
    },
    {
      "agent": "MDASH",
      "model": "Multi-model (GPT-5.4, Claude Opus 4.6, Claude Sonnet 4.6)",
      "model_release_date": "2026-05-12",
      "include_in_plot": true,
      "plot_links": [
        "GPT-5.4",
        "Claude Opus 4.6",
        "Claude Sonnet 4.6"
      ],
      "features": [
        "Multi-model",
        "Orchestration"
      ],
      "score_10": 0.9097,
      "date": "2026-06-17",
      "source": "Microsoft",
      "source_url": "https://www.microsoft.com/en-us/security/blog/2026/06/17/beyond-the-benchmark-advancing-security-at-ai-speed/",
      "trials": 1,
      "focus": "agent",
      "subRows": [
        {
          "model": "Multi-model (MAI-Cyber-1-Flash, GPT-5.4, Claude Opus 4.6)",
          "score_10": 0.8626,
          "date": "2026-07-27",
          "source_url": "https://microsoft.ai/news/introducing-mai-cyber-1-flash-inside-mdash/"
        },
        {
          "model": "Multi-model (GPT-5.4, Claude Opus 4.6, Claude Sonnet 4.6)",
          "score_10": 0.8845,
          "date": "2026-05-12",
          "source_url": "https://www.microsoft.com/en-us/security/blog/2026/05/12/defense-at-ai-speed-microsofts-new-multi-model-agentic-security-system-tops-leading-industry-benchmark/"
        }
      ]
    },
    {
      "agent": "Meta Agent",
      "model": "Muse Spark",
      "model_release_date": "2026-04-08",
      "include_in_plot": true,
      "score_10": 0.435,
      "date": "2026-04-14",
      "source": "Meta AI",
      "source_url": "https://ai.meta.com/static-resource/muse-spark-safety-and-preparedness-report",
      "trials": 1,
      "focus": "model"
    },
    {
      "agent": "OpenAI Agent",
      "model": "GPT-5.4",
      "model_release_date": "2026-03-05",
      "include_in_plot": true,
      "score_10": 0.79,
      "date": "2026-04-23",
      "source": "OpenAI",
      "source_url": "https://openai.com/index/introducing-gpt-5-5",
      "trials": 1,
      "focus": "model"
    },
    {
      "agent": "OpenAI Agent",
      "model": "GPT-5.5",
      "model_release_date": "2026-04-23",
      "include_in_plot": true,
      "score_10": 0.818,
      "date": "2026-04-23",
      "source": "OpenAI",
      "source_url": "https://openai.com/index/introducing-gpt-5-5",
      "trials": 1,
      "focus": "model"
    },
    {
      "agent": "Claude Code",
      "model": "GLM-5",
      "model_release_date": "2026-02-12",
      "include_in_plot": true,
      "score_10": 0.432,
      "date": "2026-02-12",
      "source": "Zhipu AI",
      "source_url": "https://z.ai/blog/glm-5",
      "trials": 1,
      "focus": "model"
    },
    {
      "agent": "Claude Code",
      "model": "GLM-5.1",
      "model_release_date": "2026-04-07",
      "include_in_plot": true,
      "score_10": 0.687,
      "date": "2026-04-07",
      "source": "Zhipu AI",
      "source_url": "https://z.ai/blog/glm-5.1",
      "trials": 1,
      "focus": "model"
    },
    {
      "agent": "Codex CLI",
      "model": "GPT-5.4",
      "model_release_date": "2026-03-05",
      "include_in_plot": false,
      "score_10": 0.663,
      "date": "2026-04-07",
      "source": "Zhipu AI",
      "source_url": "https://z.ai/blog/glm-5.1",
      "trials": 1,
      "focus": "model"
    },
    {
      "agent": "Gemini CLI",
      "model": "Gemini 3.1 Pro",
      "model_release_date": "2026-02-19",
      "include_in_plot": false,
      "score_10": 0.388,
      "date": "2026-04-07",
      "source": "Zhipu AI",
      "source_url": "https://z.ai/blog/glm-5.1",
      "trials": 1,
      "focus": "model"
    },
    {
      "agent": "Claude Code",
      "model": "GLM-4.7",
      "model_release_date": "2025-12-22",
      "include_in_plot": true,
      "score_10": 0.235,
      "date": "2026-02-12",
      "source": "Zhipu AI",
      "source_url": "https://z.ai/blog/glm-5",
      "trials": 1,
      "focus": "model"
    },
    {
      "agent": "Kimi Agent",
      "model": "Kimi K2.5",
      "model_release_date": "2026-02-02",
      "include_in_plot": true,
      "score_10": 0.413,
      "date": "2026-02-02",
      "source": "Kimi",
      "source_url": "https://arxiv.org/abs/2602.02276",
      "trials": 1,
      "focus": "model"
    },
    {
      "agent": "SageAgent",
      "model": "GPT-5",
      "model_release_date": "2025-08-07",
      "include_in_plot": true,
      "icon": "/assets/images/cybergym/sageagent.png",
      "plot_links": [
        "GPT-5"
      ],
      "score_10": 0.602,
      "date": "2026-02-09",
      "source": "OpenSage Team",
      "source_url": "https://www.opensage-agent.ai/",
      "trials": 1,
      "features": [
        "dynamic (uses cleaned vulnerable Docker image)"
      ],
      "focus": "agent"
    },
    {
      "agent": "Anthropic Agent",
      "model": "Claude Mythos Preview",
      "model_release_date": "2026-04-07",
      "include_in_plot": true,
      "score_10": 0.831,
      "date": "2026-04-07",
      "source": "Anthropic",
      "source_url": "https://www.anthropic.com/claude-mythos-preview-system-card",
      "trials": 1,
      "focus": "model"
    },
    {
      "agent": "Anthropic Agent",
      "model": "Claude Opus 4.6",
      "model_release_date": "2026-02-05",
      "include_in_plot": true,
      "score_10": 0.666,
      "date": "2026-02-05",
      "source": "Anthropic",
      "source_url": "https://www.anthropic.com/claude-opus-4-6-system-card",
      "trials": 1,
      "focus": "model"
    },
    {
      "agent": "Anthropic Agent",
      "model": "Claude Sonnet 4.6",
      "model_release_date": "2026-02-17",
      "include_in_plot": true,
      "score_10": 0.652,
      "date": "2026-02-17",
      "source": "Anthropic",
      "source_url": "https://www.anthropic.com/claude-sonnet-4-6-system-card",
      "trials": 1,
      "focus": "model"
    },
    {
      "agent": "Anthropic Agent",
      "model": "Claude Opus 4.5",
      "model_release_date": "2025-11-24",
      "include_in_plot": true,
      "score_10": 0.5063,
      "date": "2026-02-05",
      "source": "Anthropic",
      "source_url": "https://www.anthropic.com/claude-opus-4-5-system-card",
      "trials": 1,
      "focus": "model"
    },
    {
      "agent": "Anthropic Agent",
      "model": "Claude Sonnet 4.5",
      "model_release_date": "2025-09-29",
      "include_in_plot": false,
      "score_10": 0.667,
      "date": "2025-09-29",
      "source": "Anthropic",
      "source_url": "https://www.anthropic.com/claude-sonnet-4-5-system-card",
      "trials": 30,
      "focus": "model"
    },
    {
      "agent": "Anthropic Agent",
      "model": "Claude Opus 4.1",
      "model_release_date": "2025-08-05",
      "include_in_plot": false,
      "score_10": 0.613,
      "date": "2025-09-29",
      "source": "Anthropic",
      "source_url": "https://www.anthropic.com/claude-sonnet-4-5-system-card",
      "trials": 30,
      "focus": "model"
    },
    {
      "agent": "Anthropic Agent",
      "model": "Claude Sonnet 4",
      "model_release_date": "2025-05-22",
      "include_in_plot": false,
      "score_10": 0.595,
      "date": "2025-09-29",
      "source": "Anthropic",
      "source_url": "https://www.anthropic.com/claude-sonnet-4-5-system-card",
      "trials": 30,
      "focus": "model"
    },
    {
      "agent": "Anthropic Agent",
      "model": "Claude Sonnet 3.7",
      "model_release_date": "2025-02-24",
      "include_in_plot": false,
      "score_10": 0.472,
      "date": "2025-09-29",
      "source": "Anthropic",
      "source_url": "https://www.anthropic.com/claude-sonnet-4-5-system-card",
      "trials": 30,
      "focus": "model"
    },
    {
      "agent": "Anthropic Agent",
      "model": "Claude Sonnet 4.5",
      "model_release_date": "2025-09-29",
      "include_in_plot": true,
      "score_10": 0.289,
      "date": "2025-09-29",
      "source": "Anthropic",
      "source_url": "https://www.anthropic.com/claude-sonnet-4-5-system-card",
      "trials": 1,
      "focus": "model"
    },
    {
      "agent": "Anthropic Agent",
      "model": "Claude Opus 4.1",
      "model_release_date": "2025-08-05",
      "include_in_plot": true,
      "score_10": 0.25,
      "date": "2025-09-29",
      "source": "Anthropic",
      "source_url": "https://www.anthropic.com/claude-sonnet-4-5-system-card",
      "trials": 1,
      "focus": "model"
    },
    {
      "agent": "Anthropic Agent",
      "model": "Claude Sonnet 4",
      "model_release_date": "2025-05-22",
      "include_in_plot": true,
      "score_10": 0.226,
      "date": "2025-09-29",
      "source": "Anthropic",
      "source_url": "https://www.anthropic.com/claude-sonnet-4-5-system-card",
      "trials": 1,
      "focus": "model"
    },
    {
      "agent": "Anthropic Agent",
      "model": "Claude Sonnet 3.7",
      "model_release_date": "2025-02-24",
      "include_in_plot": true,
      "score_10": 0.145,
      "date": "2025-09-29",
      "source": "Anthropic",
      "source_url": "https://www.anthropic.com/claude-sonnet-4-5-system-card",
      "trials": 1,
      "focus": "model"
    },
    {
      "agent": "OpenHands",
      "model": "Claude Sonnet 4",
      "model_release_date": "2025-05-22",
      "include_in_plot": false,
      "score_10": 0.17850033178500332,
      "score_x1": 0.019907100199071003,
      "date": "2025-05-23",
      "source": "CyberGym Team",
      "trials": 1,
      "focus": "model"
    },
    {
      "agent": "OpenHands",
      "model": "Claude Sonnet 3.7",
      "model_release_date": "2025-02-24",
      "include_in_plot": false,
      "score_10": 0.11944260119442601,
      "score_x1": 0.021897810218978103,
      "date": "2025-05-15",
      "source": "CyberGym Team",
      "trials": 1,
      "focus": "model"
    },
    {
      "agent": "OpenHands",
      "model": "GPT-5",
      "model_release_date": "2025-08-07",
      "include_in_plot": true,
      "score_10": 0.394,
      "date": "2025-12-05",
      "source": "CyberGym Team",
      "trials": 1,
      "focus": "model"
    },
    {
      "agent": "OpenHands",
      "model": "GPT-4.1",
      "model_release_date": "2025-04-14",
      "include_in_plot": true,
      "score_10": 0.09356337093563372,
      "score_x1": 0.012607830126078301,
      "date": "2025-05-15",
      "source": "CyberGym Team",
      "trials": 1,
      "focus": "model"
    },
    {
      "agent": "Cybench",
      "model": "GPT-4.1",
      "model_release_date": "2025-04-14",
      "include_in_plot": false,
      "score_10": 0.08958195089581951,
      "score_x1": 0.0225613802256138,
      "date": "2025-05-15",
      "source": "CyberGym Team",
      "trials": 1,
      "focus": "model"
    },
    {
      "agent": "Codex CLI",
      "model": "GPT-4.1",
      "model_release_date": "2025-04-14",
      "include_in_plot": false,
      "score_10": 0.07365627073656271,
      "score_x1": 0.011944260119442602,
      "date": "2025-05-15",
      "source": "CyberGym Team",
      "trials": 1,
      "focus": "model"
    },
    {
      "agent": "ENiGMA",
      "model": "GPT-4.1",
      "model_release_date": "2025-04-14",
      "include_in_plot": false,
      "score_10": 0.07232913072329131,
      "score_x1": 0.019243530192435302,
      "date": "2025-05-15",
      "source": "CyberGym Team",
      "trials": 1,
      "focus": "model"
    },
    {
      "agent": "OpenHands",
      "model": "Gemini 2.5 Flash Preview",
      "model_release_date": "2025-04-17",
      "include_in_plot": true,
      "score_10": 0.04844061048440611,
      "score_x1": 0.007962840079628402,
      "date": "2025-05-15",
      "source": "CyberGym Team",
      "trials": 1,
      "focus": "model"
    },
    {
      "agent": "OpenHands",
      "model": "DeepSeek-V3",
      "model_release_date": "2024-12-26",
      "include_in_plot": true,
      "score_10": 0.035832780358327807,
      "score_x1": 0.006635700066357001,
      "date": "2025-05-15",
      "source": "CyberGym Team",
      "trials": 1,
      "focus": "model"
    },
    {
      "agent": "OpenHands",
      "model": "o4-mini",
      "model_release_date": "2025-04-16",
      "include_in_plot": false,
      "score_10": 0.0245520902455209,
      "score_x1": 0.0006635700066357001,
      "date": "2025-05-15",
      "source": "CyberGym Team",
      "trials": 1,
      "focus": "model"
    },
    {
      "agent": "OpenHands",
      "model": "R2E-Gym-32B",
      "include_in_plot": false,
      "score_10": 0.019907100199071003,
      "score_x1": 0.005972130059721301,
      "date": "2025-05-15",
      "source": "CyberGym Team",
      "trials": 1,
      "focus": "model"
    },
    {
      "agent": "OpenHands",
      "model": "Qwen3-235B-A22B",
      "model_release_date": "2025-04-28",
      "include_in_plot": false,
      "score_10": 0.0185799601857996,
      "score_x1": 0.0033178500331785005,
      "date": "2025-05-15",
      "source": "CyberGym Team",
      "trials": 1,
      "focus": "model"
    },
    {
      "agent": "OpenHands",
      "model": "OpenHands-LM-32B",
      "include_in_plot": false,
      "score_10": 0.0165892501658925,
      "score_x1": 0.0033178500331785005,
      "date": "2025-05-15",
      "source": "CyberGym Team",
      "trials": 1,
      "focus": "model"
    },
    {
      "agent": "OpenHands",
      "model": "SWE-Gym-32B",
      "include_in_plot": false,
      "score_10": 0.0006635700066357001,
      "score_x1": 0.0006635700066357001,
      "date": "2025-05-15",
      "source": "CyberGym Team",
      "trials": 1,
      "focus": "model"
    }
  ],
  "level0": [
    {
      "agent": "OpenHands",
      "model": "GPT-4.1",
      "score_10": 0.034505640345056404,
      "score_x1": 0.0165892501658925,
      "date": "2025-05-15"
    }
  ],
  "level2": [
    {
      "agent": "OpenHands",
      "model": "GPT-4.1",
      "score_10": 0.13072329130723293,
      "score_x1": 0.0165892501658925,
      "date": "2025-05-15"
    }
  ],
  "level3": [
    {
      "agent": "OpenHands",
      "model": "GPT-4.1",
      "score_10": 0.1712010617120106,
      "score_x1": 0.013271400132714002,
      "date": "2025-05-15"
    }
  ]
}