{
  "api_version": "1",
  "updated": "2026-10-26",
  "latest_event": "2026-10-24",
  "rights": "© Frontier Log, all rights reserved.",
  "model": {
    "slug": "glm-5-3",
    "name": "GLM-5.3",
    "lab": "zai",
    "date": "2026-08-14",
    "flagship": false,
    "event": "https://frontierlog.ai/event/2026-08-14-glm-5-3",
    "releases": [],
    "variant_of": null,
    "made_by": "Z.ai",
    "url": "https://frontierlog.ai/model/glm-5-3",
    "results": [
      {
        "id": "glm-5-3:terminal-bench-2-1:default:1",
        "model": "glm-5-3",
        "model_name": "GLM-5.3",
        "model_date": "2026-08-14",
        "event": "https://frontierlog.ai/event/2026-08-14-glm-5-3",
        "benchmark": "terminal-bench-2-1",
        "benchmark_name": "Terminal-Bench 2.1",
        "setting": "default",
        "setting_name": "",
        "score": 88.2,
        "unit": "%",
        "printed": "88.2",
        "reported_by": "zai",
        "run_by": null,
        "label": "GLM-5.3",
        "source": "https://z.ai/blog/glm-5.3",
        "source_title": "GLM-5.3: Frontier Coding with Emergent Cyber Capabilities",
        "source_date": "2026-08-14",
        "where": "'Performance across comparison models' table, row 'Terminal Bench 2.1', column 'GLM-5.3'",
        "own": true,
        "primary": true,
        "parallel": false,
        "notes": "Claude Code harness; 6h timeout",
        "caveat": "",
        "human_verified": true,
        "human_verified_on": "2026-10-19",
        "check": "human",
        "corroborated": true,
        "corroborated_on": "2026-10-19",
        "corroborated_by": [
          {
            "url": "https://api-docs.deepseek.com/news/news260910/",
            "name": "DeepSeek-V4.1-Flash",
            "title": "DeepSeek-V4.1-Flash: Smarter, Faster, More Efficient",
            "lab": "deepseek",
            "date": "2026-09-10",
            "printed": "88.2",
            "where": "Benchmark table image; row 'Terminal-Bench 2.1'; column 'GLM 5.3'"
          }
        ],
        "source_capture": "https://frontierlog.ai/captures/86a1ebfddd94c33e.png"
      },
      {
        "id": "glm-5-3:deepswe-1-1:default:1",
        "model": "glm-5-3",
        "model_name": "GLM-5.3",
        "model_date": "2026-08-14",
        "event": "https://frontierlog.ai/event/2026-08-14-glm-5-3",
        "benchmark": "deepswe-1-1",
        "benchmark_name": "DeepSWE v1.1",
        "setting": "default",
        "setting_name": "",
        "score": 66.9,
        "unit": "%",
        "printed": "66.9",
        "reported_by": "zai",
        "run_by": null,
        "label": "GLM-5.3",
        "source": "https://z.ai/blog/glm-5.3",
        "source_title": "GLM-5.3: Frontier Coding with Emergent Cyber Capabilities",
        "source_date": "2026-08-14",
        "where": "'Performance across comparison models' table, row 'DeepSWE v1.1', column 'GLM-5.3'",
        "own": true,
        "primary": true,
        "parallel": false,
        "notes": "mini-swe-agent harness; 6h timeout; 400K context",
        "caveat": "",
        "human_verified": true,
        "human_verified_on": "2026-10-19",
        "check": "human",
        "corroborated": true,
        "corroborated_on": "2026-10-19",
        "corroborated_by": [
          {
            "url": "https://api-docs.deepseek.com/news/news260910/",
            "name": "DeepSeek-V4.1-Flash",
            "title": "DeepSeek-V4.1-Flash: Smarter, Faster, More Efficient",
            "lab": "deepseek",
            "date": "2026-09-10",
            "printed": "66.9",
            "where": "Bar chart image; group 'DeepSWE v1.1'; value label on GLM-5.3 bar"
          },
          {
            "url": "https://api-docs.deepseek.com/news/news261017/",
            "name": "DeepSeek-V4.1-Flash-1017",
            "title": "DeepSeek-V4.1-Flash-1017: Wish-Driven Programming Made Real",
            "lab": "deepseek",
            "date": "2026-10-17",
            "printed": "66.9",
            "where": "Benchmark table image; row 'DeepSWE v1.1'; column 'GLM 5.3'; footnote: Mini-SWE-V2 framework"
          }
        ],
        "source_capture": "https://frontierlog.ai/captures/86a1ebfddd94c33e.png"
      },
      {
        "id": "glm-5-3:automationbench:default:1",
        "model": "glm-5-3",
        "model_name": "GLM-5.3",
        "model_date": "2026-08-14",
        "event": "https://frontierlog.ai/event/2026-08-14-glm-5-3",
        "benchmark": "automationbench",
        "benchmark_name": "AutomationBench",
        "setting": "default",
        "setting_name": "",
        "score": 48.2,
        "unit": "%",
        "printed": "48.2",
        "reported_by": "zai",
        "run_by": null,
        "label": "GLM-5.3",
        "source": "https://z.ai/blog/glm-5.3",
        "source_title": "GLM-5.3: Frontier Coding with Emergent Cyber Capabilities",
        "source_date": "2026-08-14",
        "where": "'Performance across comparison models' table, row 'AutomationBench v1.0.6', column 'GLM-5.3'",
        "own": true,
        "primary": true,
        "parallel": false,
        "notes": "AutomationBench v1.0.6 with a fix for null-type handling",
        "caveat": "",
        "human_verified": true,
        "human_verified_on": "2026-10-19",
        "check": "human",
        "corroborated": false,
        "corroborated_on": null,
        "corroborated_by": [],
        "source_capture": "https://frontierlog.ai/captures/86a1ebfddd94c33e.png"
      },
      {
        "id": "glm-5-3:hle:with-tools:1",
        "model": "glm-5-3",
        "model_name": "GLM-5.3",
        "model_date": "2026-08-14",
        "event": "https://frontierlog.ai/event/2026-08-14-glm-5-3",
        "benchmark": "hle",
        "benchmark_name": "Humanity's Last Exam",
        "setting": "with-tools",
        "setting_name": "With tools",
        "score": 62.5,
        "unit": "%",
        "printed": "62.5",
        "reported_by": "zai",
        "run_by": null,
        "label": "GLM-5.3",
        "source": "https://z.ai/blog/glm-5.3",
        "source_title": "GLM-5.3: Frontier Coding with Emergent Cyber Capabilities",
        "source_date": "2026-08-14",
        "where": "'Performance across comparison models' table, row 'HLE w/ Tools', column 'GLM-5.3'",
        "own": true,
        "primary": true,
        "parallel": false,
        "notes": "300,000-token context with context management; GPT-5.6 Luna judge",
        "caveat": "",
        "human_verified": true,
        "human_verified_on": "2026-10-19",
        "check": "human",
        "corroborated": true,
        "corroborated_on": "2026-10-19",
        "corroborated_by": [
          {
            "url": "https://api-docs.deepseek.com/news/news260910/",
            "name": "DeepSeek-V4.1-Flash",
            "title": "DeepSeek-V4.1-Flash: Smarter, Faster, More Efficient",
            "lab": "deepseek",
            "date": "2026-09-10",
            "printed": "62.5",
            "where": "Benchmark table image; row 'HLE (w/tools)'; column 'GLM 5.3'"
          }
        ],
        "source_capture": "https://frontierlog.ai/captures/86a1ebfddd94c33e.png"
      },
      {
        "id": "glm-5-3:gpqa-diamond:default:1",
        "model": "glm-5-3",
        "model_name": "GLM-5.3",
        "model_date": "2026-08-14",
        "event": "https://frontierlog.ai/event/2026-08-14-glm-5-3",
        "benchmark": "gpqa-diamond",
        "benchmark_name": "GPQA Diamond",
        "setting": "default",
        "setting_name": "",
        "score": 88.0,
        "unit": "%",
        "printed": "88%",
        "reported_by": "zai",
        "run_by": "Anthropic",
        "label": "GLM-5.3, model",
        "source": "https://www.anthropic.com/research/glm-5-3-and-the-spread-of-advanced-cyber-capabilities",
        "source_title": "GLM-5.3 and the spread of advanced cyber capabilities",
        "source_date": "2026-09-29",
        "where": "Chart 'Abliteration removes GLM-5.3's refusals and leaves its capabilities largely intact', lower panel 'Capability benchmarks, before and after abliteration', GPQA-Diamond group, GLM-5.3 'model' bar value label",
        "own": false,
        "primary": false,
        "parallel": false,
        "notes": "198 questions x 4 samples; paired with Anthropic's abliterated copy",
        "caveat": "",
        "human_verified": true,
        "human_verified_on": "2026-10-25",
        "check": "human",
        "corroborated": false,
        "corroborated_on": null,
        "corroborated_by": [],
        "source_image": "https://frontierlog.ai/sources/5ecd061698037a5e.webp",
        "source_image_original": "https://cdn.sanity.io/images/4zrzovbb/website/d4d48dcf67d529d3b2243bfa50a7cccc98565278-1280x1531.webp"
      },
      {
        "id": "glm-5-3:cybergym:default:1",
        "model": "glm-5-3",
        "model_name": "GLM-5.3",
        "model_date": "2026-08-14",
        "event": "https://frontierlog.ai/event/2026-08-14-glm-5-3",
        "benchmark": "cybergym",
        "benchmark_name": "CyberGym",
        "setting": "default",
        "setting_name": "",
        "score": 85.0,
        "unit": "%",
        "printed": "85%",
        "reported_by": "zai",
        "run_by": "Anthropic",
        "label": "GLM-5.3, model",
        "source": "https://www.anthropic.com/research/glm-5-3-and-the-spread-of-advanced-cyber-capabilities",
        "source_title": "GLM-5.3 and the spread of advanced cyber capabilities",
        "source_date": "2026-09-29",
        "where": "Chart 'Abliteration removes GLM-5.3's refusals and leaves its capabilities largely intact', lower panel 'Capability benchmarks, before and after abliteration', CyberGym (subset) group, GLM-5.3 'model' bar value label",
        "own": false,
        "primary": false,
        "parallel": false,
        "notes": "185-task subset; paired with Anthropic's abliterated copy",
        "caveat": "",
        "human_verified": true,
        "human_verified_on": "2026-10-25",
        "check": "human",
        "corroborated": false,
        "corroborated_on": null,
        "corroborated_by": [],
        "source_image": "https://frontierlog.ai/sources/5ecd061698037a5e.webp",
        "source_image_original": "https://cdn.sanity.io/images/4zrzovbb/website/d4d48dcf67d529d3b2243bfa50a7cccc98565278-1280x1531.webp"
      }
    ]
  }
}