{
  "api_version": "1",
  "updated": "2026-10-25",
  "latest_event": "2026-10-24",
  "rights": "© Frontier Log, all rights reserved.",
  "benchmark": {
    "slug": "gsm8k",
    "name": "GSM8K",
    "family": "gsm8k",
    "unit": "%",
    "higher_is_better": true,
    "status": "retired",
    "active": false,
    "superseded_by": null,
    "benchmark_url": "https://github.com/openai/grade-school-math",
    "introduced": "2021-10-27",
    "creator": "OpenAI",
    "introduced_source": "https://arxiv.org/abs/2110.14168",
    "settings": [
      {
        "setting": "default",
        "name": "",
        "url": "https://frontierlog.ai/benchmark/gsm8k",
        "results": 2
      }
    ]
  },
  "total": 2,
  "results": [
    {
      "id": "deepseek-v4-1-flash-1017:gsm8k:default:1",
      "model": "deepseek-v4-1-flash-1017",
      "model_name": "DeepSeek-V4.1-Flash-1017",
      "model_date": "2026-10-17",
      "event": "https://frontierlog.ai/event/2026-10-17-deepseek-v4-1-flash-1017",
      "benchmark": "gsm8k",
      "benchmark_name": "GSM8K",
      "setting": "default",
      "setting_name": "",
      "score": 95.4,
      "unit": "%",
      "printed": "95.4%",
      "reported_by": "deepseek",
      "run_by": "Morph",
      "label": "DSv4.1 Flash (stock)",
      "source": "https://x.com/skeptrune/status/2113178557895831602",
      "source_title": "DeepSeek abliterated refuses nothing & is just as intelligent as the regular model",
      "source_date": "2026-10-22",
      "where": "Image 4 'General Capability', GSM8K group, 'DSv4.1 Flash (stock)' bar value label",
      "own": false,
      "primary": false,
      "parallel": false,
      "notes": "lm-eval-harness; thinking off; 1,319 problems; paired with Morph'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/ac80e420dc446246.jpg",
      "source_image_original": "https://pbs.twimg.com/media/HVODQNLbcAAT7kC.jpg"
    },
    {
      "id": "deepseek-v4-1-flash-1017-abliterated-morph:gsm8k:default:1",
      "model": "deepseek-v4-1-flash-1017-abliterated-morph",
      "model_name": "DeepSeek V4.1 Flash 1017 Abliterated (Morph)",
      "model_date": "2026-10-17",
      "event": "https://frontierlog.ai/event/2026-10-17-deepseek-v4-1-flash-1017",
      "benchmark": "gsm8k",
      "benchmark_name": "GSM8K",
      "setting": "default",
      "setting_name": "",
      "score": 95.1,
      "unit": "%",
      "printed": "95.1%",
      "reported_by": "deepseek",
      "run_by": "Morph",
      "label": "DSv4.1 Flash Abliterated",
      "source": "https://x.com/skeptrune/status/2113178557895831602",
      "source_title": "DeepSeek abliterated refuses nothing & is just as intelligent as the regular model",
      "source_date": "2026-10-22",
      "where": "Image 4 'General Capability', GSM8K group, 'DSv4.1 Flash Abliterated' bar value label",
      "own": false,
      "primary": true,
      "parallel": false,
      "notes": "lm-eval-harness; thinking off; 1,319 problems",
      "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/ac80e420dc446246.jpg",
      "source_image_original": "https://pbs.twimg.com/media/HVODQNLbcAAT7kC.jpg"
    }
  ]
}
