{
  "question_id": "W283",
  "slug": "no-new-capability-record-for-two-years",
  "title": "Two years pass with no new record on Epoch's Capabilities Index",
  "type": "binary",
  "resolution": {
    "levels": [],
    "values": [],
    "annotations": []
  },
  "forecasts": [
    {
      "question_id": "W283",
      "forecast_date": "2026-09-20",
      "evidence_cutoff": "2026-09-20",
      "method": "wide-v2",
      "opened": "2026-09-16",
      "linked_questions": [],
      "reading_adopted": null,
      "summary": null,
      "ambiguities": [],
      "key_drivers": [],
      "would_update_on": [],
      "forecasts": [
        {
          "period_end": "2026-09-30",
          "p_yes": 0.00001
        },
        {
          "period_end": "2026-12-31",
          "p_yes": 0.00002
        },
        {
          "period_end": "2027-03-31",
          "p_yes": 0.00003
        },
        {
          "period_end": "2027-06-30",
          "p_yes": 0.00005
        },
        {
          "period_end": "2027-09-30",
          "p_yes": 0.00008
        },
        {
          "period_end": "2027-12-31",
          "p_yes": 0.00012
        },
        {
          "period_end": "2028-03-31",
          "p_yes": 0.0002
        },
        {
          "period_end": "2028-06-30",
          "p_yes": 0.0005
        },
        {
          "period_end": "2028-09-30",
          "p_yes": 0.002499
        },
        {
          "period_end": "2028-12-31",
          "p_yes": 0.005491
        },
        {
          "period_end": "2029-03-31",
          "p_yes": 0.008972
        },
        {
          "period_end": "2029-06-30",
          "p_yes": 0.012936
        },
        {
          "period_end": "2029-09-30",
          "p_yes": 0.017378
        },
        {
          "period_end": "2029-12-31",
          "p_yes": 0.022291
        },
        {
          "period_end": "2030-03-31",
          "p_yes": 0.027669
        },
        {
          "period_end": "2030-06-30",
          "p_yes": 0.033503
        },
        {
          "period_end": "2030-09-30",
          "p_yes": 0.039785
        },
        {
          "period_end": "2030-12-31",
          "p_yes": 0.046506
        },
        {
          "period_end": "2031-03-31",
          "p_yes": 0.053658
        },
        {
          "period_end": "2031-06-30",
          "p_yes": 0.061228
        },
        {
          "period_end": "2031-09-30",
          "p_yes": 0.069208
        }
      ],
      "passes": []
    },
    {
      "question_id": "W283",
      "forecast_date": "2026-09-27",
      "evidence_cutoff": "2026-09-27",
      "method": "wide-v2",
      "opened": "2026-09-16",
      "linked_questions": [],
      "reading_adopted": null,
      "summary": null,
      "ambiguities": [],
      "key_drivers": [],
      "would_update_on": [],
      "forecasts": [
        {
          "period_end": "2026-09-30",
          "p_yes": 0.00001
        },
        {
          "period_end": "2026-12-31",
          "p_yes": 0.00002
        },
        {
          "period_end": "2027-03-31",
          "p_yes": 0.00004
        },
        {
          "period_end": "2027-06-30",
          "p_yes": 0.00006
        },
        {
          "period_end": "2027-09-30",
          "p_yes": 0.00008
        },
        {
          "period_end": "2027-12-31",
          "p_yes": 0.0001
        },
        {
          "period_end": "2028-03-31",
          "p_yes": 0.00015
        },
        {
          "period_end": "2028-06-30",
          "p_yes": 0.00025
        },
        {
          "period_end": "2028-09-30",
          "p_yes": 0.00175
        },
        {
          "period_end": "2028-12-31",
          "p_yes": 0.00425
        },
        {
          "period_end": "2029-03-31",
          "p_yes": 0.00773
        },
        {
          "period_end": "2029-06-30",
          "p_yes": 0.0117
        },
        {
          "period_end": "2029-09-30",
          "p_yes": 0.01615
        },
        {
          "period_end": "2029-12-31",
          "p_yes": 0.02107
        },
        {
          "period_end": "2030-03-31",
          "p_yes": 0.02645
        },
        {
          "period_end": "2030-06-30",
          "p_yes": 0.03229
        },
        {
          "period_end": "2030-09-30",
          "p_yes": 0.03858
        },
        {
          "period_end": "2030-12-31",
          "p_yes": 0.04531
        },
        {
          "period_end": "2031-03-31",
          "p_yes": 0.05247
        },
        {
          "period_end": "2031-06-30",
          "p_yes": 0.06005
        },
        {
          "period_end": "2031-09-30",
          "p_yes": 0.06804
        }
      ],
      "passes": []
    },
    {
      "question_id": "W283",
      "forecast_date": "2026-09-30",
      "evidence_cutoff": "2026-09-30",
      "method": "wide-v2.1",
      "opened": "2026-09-16",
      "linked_questions": [],
      "reading_adopted": "Records are computed within a single Epoch data vintage by release date; with the current record at GPT-6 Astra (2026-09-03), the earliest qualifying interval end T is ~2028-09-03, and YES requires an Epoch data update/analysis dated after T showing no strictly higher model released in (S, T].",
      "summary": null,
      "ambiguities": [],
      "key_drivers": [],
      "would_update_on": [],
      "forecasts": [
        {
          "period_end": "2026-09-30",
          "p_yes": 0.001
        },
        {
          "period_end": "2026-12-31",
          "p_yes": 0.001
        },
        {
          "period_end": "2027-03-31",
          "p_yes": 0.001
        },
        {
          "period_end": "2027-06-30",
          "p_yes": 0.001
        },
        {
          "period_end": "2027-09-30",
          "p_yes": 0.001
        },
        {
          "period_end": "2027-12-31",
          "p_yes": 0.001
        },
        {
          "period_end": "2028-03-31",
          "p_yes": 0.001
        },
        {
          "period_end": "2028-06-30",
          "p_yes": 0.001
        },
        {
          "period_end": "2028-09-30",
          "p_yes": 0.002
        },
        {
          "period_end": "2028-12-31",
          "p_yes": 0.0033
        },
        {
          "period_end": "2029-03-31",
          "p_yes": 0.0048
        },
        {
          "period_end": "2029-06-30",
          "p_yes": 0.0066
        },
        {
          "period_end": "2029-09-30",
          "p_yes": 0.0087
        },
        {
          "period_end": "2029-12-31",
          "p_yes": 0.0109
        },
        {
          "period_end": "2030-03-31",
          "p_yes": 0.0133
        },
        {
          "period_end": "2030-06-30",
          "p_yes": 0.0159
        },
        {
          "period_end": "2030-09-30",
          "p_yes": 0.0187
        },
        {
          "period_end": "2030-12-31",
          "p_yes": 0.0217
        },
        {
          "period_end": "2031-03-31",
          "p_yes": 0.0246
        },
        {
          "period_end": "2031-06-30",
          "p_yes": 0.0276
        },
        {
          "period_end": "2031-09-30",
          "p_yes": 0.0306
        }
      ],
      "passes": []
    },
    {
      "question_id": "W283",
      "forecast_date": "2026-10-07",
      "evidence_cutoff": "2026-10-07",
      "method": "wide-v2.1",
      "opened": "2026-09-16",
      "linked_questions": [],
      "reading_adopted": null,
      "summary": "The top score on Epoch's index has risen every few months since 2024, most recently on September 22, 2026. So a two-year gap could not end before late 2028. For one to happen, every developer worldwide would have to stop releasing better models for two years, including Chinese labs that trail by about six months, and so far neither Washington nor Beijing supports stopping. The development that would most change this is a binding US–China freeze, likely after a serious AI incident.",
      "ambiguities": [],
      "key_drivers": [],
      "would_update_on": [],
      "forecasts": [
        {
          "period_end": "2026-12-31",
          "p_yes": 0
        },
        {
          "period_end": "2027-03-31",
          "p_yes": 0
        },
        {
          "period_end": "2027-06-30",
          "p_yes": 0
        },
        {
          "period_end": "2027-09-30",
          "p_yes": 0
        },
        {
          "period_end": "2027-12-31",
          "p_yes": 0
        },
        {
          "period_end": "2028-03-31",
          "p_yes": 0
        },
        {
          "period_end": "2028-06-30",
          "p_yes": 0
        },
        {
          "period_end": "2028-09-30",
          "p_yes": 0.0002
        },
        {
          "period_end": "2028-12-31",
          "p_yes": 0.0007
        },
        {
          "period_end": "2029-03-31",
          "p_yes": 0.0025
        },
        {
          "period_end": "2029-06-30",
          "p_yes": 0.0051
        },
        {
          "period_end": "2029-09-30",
          "p_yes": 0.0077
        },
        {
          "period_end": "2029-12-31",
          "p_yes": 0.0102
        },
        {
          "period_end": "2030-03-31",
          "p_yes": 0.013
        },
        {
          "period_end": "2030-06-30",
          "p_yes": 0.016
        },
        {
          "period_end": "2030-09-30",
          "p_yes": 0.019
        },
        {
          "period_end": "2030-12-31",
          "p_yes": 0.0219
        },
        {
          "period_end": "2031-03-31",
          "p_yes": 0.0249
        },
        {
          "period_end": "2031-06-30",
          "p_yes": 0.0278
        },
        {
          "period_end": "2031-09-30",
          "p_yes": 0.0308
        },
        {
          "period_end": "2031-12-31",
          "p_yes": 0.0338
        }
      ],
      "passes": []
    },
    {
      "question_id": "W283",
      "forecast_date": "2026-10-14",
      "evidence_cutoff": "2026-10-14",
      "method": "wide-v2.1",
      "opened": "2026-09-16",
      "linked_questions": [],
      "reading_adopted": null,
      "summary": "Epoch's capability index has hit a new high every one to three months. Its longest gap ever is about ten and a half months, and the latest high was set on September 22, 2026. A yes needs two full years in which no model from any developer, Chinese labs included, beats the top score, which would basically take a binding global pause. The development that would most change this forecast is a binding US–China agreement halting frontier AI releases.",
      "ambiguities": [],
      "key_drivers": [],
      "would_update_on": [],
      "forecasts": [
        {
          "period_end": "2026-12-31",
          "p_yes": 0
        },
        {
          "period_end": "2027-03-31",
          "p_yes": 0
        },
        {
          "period_end": "2027-06-30",
          "p_yes": 0
        },
        {
          "period_end": "2027-09-30",
          "p_yes": 0
        },
        {
          "period_end": "2027-12-31",
          "p_yes": 0
        },
        {
          "period_end": "2028-03-31",
          "p_yes": 0
        },
        {
          "period_end": "2028-06-30",
          "p_yes": 0
        },
        {
          "period_end": "2028-09-30",
          "p_yes": 0.0002
        },
        {
          "period_end": "2028-12-31",
          "p_yes": 0.0008
        },
        {
          "period_end": "2029-03-31",
          "p_yes": 0.0018
        },
        {
          "period_end": "2029-06-30",
          "p_yes": 0.0032
        },
        {
          "period_end": "2029-09-30",
          "p_yes": 0.0048
        },
        {
          "period_end": "2029-12-31",
          "p_yes": 0.0066
        },
        {
          "period_end": "2030-03-31",
          "p_yes": 0.0084
        },
        {
          "period_end": "2030-06-30",
          "p_yes": 0.0102
        },
        {
          "period_end": "2030-09-30",
          "p_yes": 0.0121
        },
        {
          "period_end": "2030-12-31",
          "p_yes": 0.0142
        },
        {
          "period_end": "2031-03-31",
          "p_yes": 0.0169
        },
        {
          "period_end": "2031-06-30",
          "p_yes": 0.0199
        },
        {
          "period_end": "2031-09-30",
          "p_yes": 0.0231
        },
        {
          "period_end": "2031-12-31",
          "p_yes": 0.0263
        }
      ],
      "passes": []
    }
  ],
  "declined": []
}