{
  "api_version": "1",
  "updated": "2026-10-26",
  "latest_event": "2026-10-24",
  "rights": "© Frontier Log, all rights reserved.",
  "model": {
    "slug": "deepseek-v3-2",
    "name": "DeepSeek-V3.2",
    "lab": "deepseek",
    "date": "2025-12-01",
    "flagship": false,
    "event": "https://frontierlog.ai/event/2025-12-01-deepseek-v3-2",
    "releases": [],
    "variant_of": null,
    "made_by": "DeepSeek",
    "url": "https://frontierlog.ai/model/deepseek-v3-2",
    "results": [
      {
        "id": "deepseek-v3-2:swe-bench-verified:default:1",
        "model": "deepseek-v3-2",
        "model_name": "DeepSeek-V3.2",
        "model_date": "2025-12-01",
        "event": "https://frontierlog.ai/event/2025-12-01-deepseek-v3-2",
        "benchmark": "swe-bench-verified",
        "benchmark_name": "SWE-bench Verified",
        "setting": "default",
        "setting_name": "",
        "score": 73.1,
        "unit": "%",
        "printed": "73.1",
        "reported_by": "deepseek",
        "run_by": null,
        "label": "DeepSeek-V3.2-Thinking",
        "source": "https://api-docs.deepseek.com/news/news251201/",
        "source_title": "DeepSeek-V3.2 Release",
        "source_date": "2025-12-01",
        "where": "benchmark chart image, 'SWE Verified (Resolved)', bar 'DeepSeek-V3.2-Thinking'",
        "own": true,
        "primary": true,
        "parallel": false,
        "notes": "thinking mode",
        "caveat": "",
        "human_verified": true,
        "human_verified_on": "2026-10-17",
        "check": "human",
        "corroborated": true,
        "corroborated_on": "2026-10-19",
        "corroborated_by": [
          {
            "url": "https://www.minimax.io/news/minimax-m21",
            "name": "MiniMax M2.1",
            "title": "MiniMax M2.1: Significantly Enhanced Multi-Language Programming, Built for Real-World Complex Tasks",
            "lab": "minimax",
            "date": "2025-12-23",
            "printed": "73.1",
            "where": "Coding bar chart image (10 panels); panel 'SWE-bench Verified'; value label on DeepSeek V3.2 bar"
          },
          {
            "url": "https://www.kimi.ai/blog/kimi-k2-5",
            "name": "Kimi K2.5",
            "title": "Kimi K2.5: Visual Agentic Intelligence",
            "lab": "moonshot",
            "date": "2026-01-27",
            "printed": "73.1",
            "where": "Appendix Benchmark table, row SWE-Bench Verified ; column DeepSeek V3.2 (Thinking)"
          },
          {
            "url": "https://qwen.ai/blog?id=qwen3-max-thinking",
            "name": "Qwen3-Max-Thinking",
            "title": "Pushing Qwen3-Max-Thinking Beyond its Limits",
            "lab": "alibaba",
            "date": "2026-01-26",
            "printed": "73.1",
            "where": "Score chart image (score.png), panel SWE-bench Verified, value label on the DeepSeek-V3.2 bar"
          },
          {
            "url": "https://mistral.ai/news/devstral-2-vibe-cli/",
            "name": "Devstral 2",
            "title": "Introducing: Devstral 2 and Mistral Vibe CLI.",
            "lab": "mistral",
            "date": "2025-12-09",
            "printed": "73.1",
            "where": "Chart 'SWE-Bench Verified: Open-weight vs Proprietary models', bar 'Deepseek V3.2'"
          },
          {
            "url": "https://z.ai/blog/glm-4.7",
            "name": "GLM-4.7",
            "title": "GLM-4.7: Advancing the Coding Capability",
            "lab": "zai",
            "date": "2025-12-22",
            "printed": "73.1",
            "where": "Benchmark table, row 'SWE-bench Verified', column 'DeepSeek-V3.2'"
          },
          {
            "url": "https://z.ai/blog/glm-5",
            "name": "GLM-5",
            "title": "GLM-5: From Vibe Coding to Agentic Engineering",
            "lab": "zai",
            "date": "2026-02-12",
            "printed": "73.1",
            "where": "Benchmark table, row 'SWE-bench Verified', column 'DeepSeek-V3.2 (Thinking)'"
          }
        ],
        "source_image": "https://frontierlog.ai/sources/8ed4c04255cbd56a.webp",
        "source_image_original": "https://api-docs.deepseek.com/img/v3.2_251201_benchmark.webp"
      },
      {
        "id": "deepseek-v3-2:hle:no-tools:1",
        "model": "deepseek-v3-2",
        "model_name": "DeepSeek-V3.2",
        "model_date": "2025-12-01",
        "event": "https://frontierlog.ai/event/2025-12-01-deepseek-v3-2",
        "benchmark": "hle",
        "benchmark_name": "Humanity's Last Exam",
        "setting": "no-tools",
        "setting_name": "No tools",
        "score": 25.1,
        "unit": "%",
        "printed": "25.1",
        "reported_by": "deepseek",
        "run_by": null,
        "label": "DeepSeek-V3.2-Thinking",
        "source": "https://api-docs.deepseek.com/news/news251201/",
        "source_title": "DeepSeek-V3.2 Release",
        "source_date": "2025-12-01",
        "where": "benchmark chart image (first image), group 'HLE (Pass@1)', bar 'DeepSeek-V3.2-Thinking', value label",
        "own": true,
        "primary": true,
        "parallel": false,
        "notes": "thinking mode; 21K output tokens; full or text-only set not stated",
        "caveat": "",
        "human_verified": true,
        "human_verified_on": "2026-10-17",
        "check": "human",
        "corroborated": true,
        "corroborated_on": "2026-10-19",
        "corroborated_by": [
          {
            "url": "https://qwen.ai/blog?id=qwen3-max-thinking",
            "name": "Qwen3-Max-Thinking",
            "title": "Pushing Qwen3-Max-Thinking Beyond its Limits",
            "lab": "alibaba",
            "date": "2026-01-26",
            "printed": "25.1",
            "where": "Score chart image (score.png), panel Humanity's Last Exam, value label on the DeepSeek-V3.2 bar"
          },
          {
            "url": "https://z.ai/blog/glm-4.7",
            "name": "GLM-4.7",
            "title": "GLM-4.7: Advancing the Coding Capability",
            "lab": "zai",
            "date": "2025-12-22",
            "printed": "25.1",
            "where": "Benchmark table, row 'HLE (separate from 'HLE w/ Tools' row)', column 'DeepSeek-V3.2'"
          }
        ],
        "source_image": "https://frontierlog.ai/sources/8ed4c04255cbd56a.webp",
        "source_image_original": "https://api-docs.deepseek.com/img/v3.2_251201_benchmark.webp"
      },
      {
        "id": "deepseek-v3-2:terminal-bench-2-0:default:1",
        "model": "deepseek-v3-2",
        "model_name": "DeepSeek-V3.2",
        "model_date": "2025-12-01",
        "event": "https://frontierlog.ai/event/2025-12-01-deepseek-v3-2",
        "benchmark": "terminal-bench-2-0",
        "benchmark_name": "Terminal-Bench 2.0",
        "setting": "default",
        "setting_name": "",
        "score": 46.4,
        "unit": "%",
        "printed": "46.4",
        "reported_by": "deepseek",
        "run_by": null,
        "label": "DeepSeek-V3.2-Thinking",
        "source": "https://api-docs.deepseek.com/news/news251201/",
        "source_title": "DeepSeek-V3.2 Release",
        "source_date": "2025-12-01",
        "where": "bar chart image (v3.2_251201_benchmark), 'Agentic Capabilities' group 'Terminal Bench 2.0 (Acc)', bar 'DeepSeek-V3.2-Thinking', value label",
        "own": true,
        "primary": true,
        "parallel": false,
        "notes": "",
        "caveat": "",
        "human_verified": true,
        "human_verified_on": "2026-10-18",
        "check": "human",
        "corroborated": true,
        "corroborated_on": "2026-10-19",
        "corroborated_by": [
          {
            "url": "https://www.minimax.io/news/minimax-m21",
            "name": "MiniMax M2.1",
            "title": "MiniMax M2.1: Significantly Enhanced Multi-Language Programming, Built for Real-World Complex Tasks",
            "lab": "minimax",
            "date": "2025-12-23",
            "printed": "46.4",
            "where": "Coding bar chart image (10 panels); panel 'Terminal-bench 2.0'; value label on DeepSeek V3.2 bar"
          },
          {
            "url": "https://www.kimi.ai/blog/kimi-k2-5",
            "name": "Kimi K2.5",
            "title": "Kimi K2.5: Visual Agentic Intelligence",
            "lab": "moonshot",
            "date": "2026-01-27",
            "printed": "46.4",
            "where": "Appendix Benchmark table, row Terminal-Bench 2.0 (Terminus-2; K2.5 non-thinking mode) ; column DeepSeek V3.2 (Thinking)"
          },
          {
            "url": "https://z.ai/blog/glm-4.7",
            "name": "GLM-4.7",
            "title": "GLM-4.7: Advancing the Coding Capability",
            "lab": "zai",
            "date": "2025-12-22",
            "printed": "46.4",
            "where": "Benchmark table, row 'Terminal Bench 2.0', column 'DeepSeek-V3.2'"
          },
          {
            "url": "https://z.ai/blog/glm-5",
            "name": "GLM-5",
            "title": "GLM-5: From Vibe Coding to Agentic Engineering",
            "lab": "zai",
            "date": "2026-02-12",
            "printed": "46.4",
            "where": "Benchmark table, row 'Terminal-Bench 2.0 Claude Code (GLM-5 cell '56.2 / 61.1†': first value; 61.1 is the verified variant)', column 'DeepSeek-V3.2 (Thinking)'"
          },
          {
            "url": "https://z.ai/blog/glm-5.1",
            "name": "GLM-5.1",
            "title": "GLM-5.1: Towards Long-Horizon Tasks",
            "lab": "zai",
            "date": "2026-04-07",
            "printed": "46.4",
            "where": "Benchmark table, row 'Terminal-Bench 2.0 Best self-reported harness', column 'DeepSeek-V3.2' (Claude Code)"
          }
        ],
        "source_image": "https://frontierlog.ai/sources/8ed4c04255cbd56a.webp",
        "source_image_original": "https://api-docs.deepseek.com/img/v3.2_251201_benchmark.webp"
      },
      {
        "id": "deepseek-v3-2:aime-2025:no-tools:1",
        "model": "deepseek-v3-2",
        "model_name": "DeepSeek-V3.2",
        "model_date": "2025-12-01",
        "event": "https://frontierlog.ai/event/2025-12-01-deepseek-v3-2",
        "benchmark": "aime-2025",
        "benchmark_name": "AIME 2025",
        "setting": "no-tools",
        "setting_name": "No tools",
        "score": 93.1,
        "unit": "%",
        "printed": "93.1",
        "reported_by": "deepseek",
        "run_by": null,
        "label": "DeepSeek-V3.2 Thinking",
        "source": "https://api-docs.deepseek.com/news/news251201/",
        "source_title": "DeepSeek-V3.2 Release",
        "source_date": "2025-12-01",
        "where": "reasoning table image (v3.2_251201_benchmark_table_en), row 'AIME 2025', column 'DeepSeek-V3.2 Thinking'",
        "own": true,
        "primary": true,
        "parallel": false,
        "notes": "16K output tokens",
        "caveat": "",
        "human_verified": true,
        "human_verified_on": "2026-10-18",
        "check": "human",
        "corroborated": false,
        "corroborated_on": null,
        "corroborated_by": [],
        "source_image": "https://frontierlog.ai/sources/a313f1f47e0fe2c0.png",
        "source_image_original": "https://api-docs.deepseek.com/img/v3.2_251201_benchmark_table_en.png"
      },
      {
        "id": "deepseek-v3-2:gpqa-diamond:default:1",
        "model": "deepseek-v3-2",
        "model_name": "DeepSeek-V3.2",
        "model_date": "2025-12-01",
        "event": "https://frontierlog.ai/event/2025-12-01-deepseek-v3-2",
        "benchmark": "gpqa-diamond",
        "benchmark_name": "GPQA Diamond",
        "setting": "default",
        "setting_name": "",
        "score": 82.4,
        "unit": "%",
        "printed": "82.4",
        "reported_by": "deepseek",
        "run_by": null,
        "label": "DeepSeek-V3.2 Thinking",
        "source": "https://api-docs.deepseek.com/news/news251201/",
        "source_title": "DeepSeek-V3.2 Release",
        "source_date": "2025-12-01",
        "where": "reasoning table image (v3.2_251201_benchmark_table_en), row 'GPQA Diamond', column 'DeepSeek-V3.2 Thinking', score before the bracketed token count",
        "own": true,
        "primary": true,
        "parallel": false,
        "notes": "thinking mode; 7K output tokens",
        "caveat": "",
        "human_verified": false,
        "human_verified_on": null,
        "check": "corroborated",
        "corroborated": true,
        "corroborated_on": "2026-10-19",
        "corroborated_by": [
          {
            "url": "https://huggingface.co/deepseek-ai/DeepSeek-V3.2/resolve/main/assets/paper.pdf",
            "name": "DeepSeek-V3.2 card",
            "title": "DeepSeek-V3.2: Pushing the Frontier of Open Large Language Models",
            "lab": "deepseek",
            "date": "2025-12-01",
            "printed": "82.4",
            "where": "Table 3, row 'GPQA Diamond (Pass@1)', column 'DeepSeek-V3.2 Thinking', printed '82.4 (7k)' (also Table 2, same value)"
          }
        ],
        "source_image": "https://frontierlog.ai/sources/a313f1f47e0fe2c0.png",
        "source_image_original": "https://api-docs.deepseek.com/img/v3.2_251201_benchmark_table_en.png"
      }
    ]
  }
}