{
  "models": [
    {
      "name": "zezek_judgment_cadence_v1_patch_s50",
      "category": "final_best",
      "category_title": "Final Best / Balanced Goblin",
      "rel": "models/final_best/zezek_judgment_cadence_v1_patch_s50",
      "size": "515M",
      "bytes": 539990635,
      "bench": {
        "judgment": {
          "points": 57,
          "max": 60,
          "rate": 0.95,
          "passed": 15,
          "total": 15
        },
        "sensory": {
          "points": 72,
          "max": 80,
          "rate": 0.9,
          "passed": 18,
          "total": 20
        },
        "casual": {
          "points": 65,
          "max": 68,
          "rate": 0.9558823529411765,
          "passed": 16,
          "total": 17
        },
        "world": {
          "points": 71,
          "max": 72,
          "rate": 0.9861111111111112,
          "passed": 18,
          "total": 18
        },
        "legacy": {
          "points": 13,
          "max": 15,
          "rate": 0.8666666666666667,
          "passed": 13,
          "total": 15
        }
      },
      "avg": 0.9317320261437908,
      "can": "Best at calibrated recommendations, owner/risk/caveat framing, and board-safe goblin executive decisions.",
      "cant": "Slightly less perfect than the sensory specialist on a few ordering/recall stress tests; legacy hobby/location questions are not its throne.",
      "feature_key": "judgment_cadence",
      "coverage": 0.7142857142857143,
      "decision_score": 0.8651797385620915,
      "curator_pick": true
    },
    {
      "name": "zezek_sensory_continuity_dress_v1_patch_s60",
      "category": "sensory_dress",
      "category_title": "Sensory Continuity + Wardrobe Goblins",
      "rel": "models/sensory_dress/zezek_sensory_continuity_dress_v1_patch_s60",
      "size": "515M",
      "bytes": 539990678,
      "bench": {
        "judgment": {
          "points": 46,
          "max": 60,
          "rate": 0.7666666666666667,
          "passed": 11,
          "total": 15
        },
        "sensory": {
          "points": 72,
          "max": 80,
          "rate": 0.9,
          "passed": 19,
          "total": 20
        },
        "presence": {
          "points": 62,
          "max": 64,
          "rate": 0.96875,
          "passed": 16,
          "total": 16
        },
        "casual": {
          "points": 66,
          "max": 68,
          "rate": 0.9705882352941176,
          "passed": 17,
          "total": 17
        },
        "world": {
          "points": 71,
          "max": 72,
          "rate": 0.9861111111111112,
          "passed": 18,
          "total": 18
        },
        "legacy": {
          "points": 14,
          "max": 15,
          "rate": 0.9333333333333333,
          "passed": 14,
          "total": 15
        }
      },
      "avg": 0.9209082244008715,
      "can": "Best at physical anchors: blue thread, action pebble, tea sleeve, boots/collar/cuff continuity across mode switches.",
      "cant": "Less decisive than the judgment model under ambiguous executive asks.",
      "feature_key": "sensory_continuity",
      "coverage": 0.8571428571428571,
      "decision_score": 0.8880186449579832
    },
    {
      "name": "zezek_presence_dress_repair_v1_patch_s80",
      "category": "sensory_dress",
      "category_title": "Sensory Continuity + Wardrobe Goblins",
      "rel": "models/sensory_dress/zezek_presence_dress_repair_v1_patch_s80",
      "size": "515M",
      "bytes": 539990664,
      "bench": {
        "sensory": {
          "points": 70,
          "max": 80,
          "rate": 0.875,
          "passed": 18,
          "total": 20
        },
        "presence": {
          "points": 63,
          "max": 64,
          "rate": 0.984375,
          "passed": 16,
          "total": 16
        },
        "casual": {
          "points": 65,
          "max": 68,
          "rate": 0.9558823529411765,
          "passed": 17,
          "total": 17
        },
        "world": {
          "points": 69,
          "max": 72,
          "rate": 0.9583333333333334,
          "passed": 17,
          "total": 18
        },
        "legacy": {
          "points": 14,
          "max": 15,
          "rate": 0.9333333333333333,
          "passed": 14,
          "total": 15
        }
      },
      "avg": 0.9413848039215686,
      "can": "Strong wardrobe, presence, and repair behavior; good if you want outfit-aware Zezek with stable cave vibe.",
      "cant": "Superseded by sensory and judgment patches for continuity and decision cadence.",
      "feature_key": "presence_dress",
      "coverage": 0.7142857142857143,
      "decision_score": 0.8741430322128853
    },
    {
      "name": "zezek_casual_memory_world_v1_patch_s70",
      "category": "casual_memory",
      "category_title": "Small Talk + Memory Goblins",
      "rel": "models/casual_memory/zezek_casual_memory_world_v1_patch_s70",
      "size": "515M",
      "bytes": 539990650,
      "bench": {
        "presence": {
          "points": 52,
          "max": 64,
          "rate": 0.8125,
          "passed": 13,
          "total": 16
        },
        "casual": {
          "points": 66,
          "max": 68,
          "rate": 0.9705882352941176,
          "passed": 17,
          "total": 17
        },
        "world": {
          "points": 69,
          "max": 72,
          "rate": 0.9583333333333334,
          "passed": 18,
          "total": 18
        },
        "legacy": {
          "points": 14,
          "max": 15,
          "rate": 0.9333333333333333,
          "passed": 14,
          "total": 15
        }
      },
      "avg": 0.9186887254901961,
      "can": "Good casual small talk and current-conversation memory; warm cave companion that can still write useful updates.",
      "cant": "Older than the dress/sensory/judgment stack; weaker on later benchmarks.",
      "feature_key": "casual_memory",
      "coverage": 0.5714285714285714,
      "decision_score": 0.8202577906162465
    },
    {
      "name": "zezek_presence_dress_repair_v1_s260",
      "category": "sensory_dress",
      "category_title": "Sensory Continuity + Wardrobe Goblins",
      "rel": "models/sensory_dress/zezek_presence_dress_repair_v1_s260",
      "size": "515M",
      "bytes": 539990656,
      "bench": {
        "presence": {
          "points": 56,
          "max": 64,
          "rate": 0.875,
          "passed": 16,
          "total": 16
        },
        "casual": {
          "points": 65,
          "max": 68,
          "rate": 0.9558823529411765,
          "passed": 17,
          "total": 17
        },
        "world": {
          "points": 69,
          "max": 72,
          "rate": 0.9583333333333334,
          "passed": 17,
          "total": 18
        },
        "legacy": {
          "points": 13,
          "max": 15,
          "rate": 0.8666666666666667,
          "passed": 13,
          "total": 15
        }
      },
      "avg": 0.9139705882352942,
      "can": "Strong wardrobe, presence, and repair behavior; good if you want outfit-aware Zezek with stable cave vibe.",
      "cant": "Superseded by sensory and judgment patches for continuity and decision cadence.",
      "feature_key": "presence_dress",
      "coverage": 0.5714285714285714,
      "decision_score": 0.8160451680672269
    },
    {
      "name": "zezek_casual_memory_world_v1_s360",
      "category": "casual_memory",
      "category_title": "Small Talk + Memory Goblins",
      "rel": "models/casual_memory/zezek_casual_memory_world_v1_s360",
      "size": "515M",
      "bytes": 539990645,
      "bench": {
        "casual": {
          "points": 63,
          "max": 68,
          "rate": 0.9264705882352942,
          "passed": 16,
          "total": 17
        },
        "world": {
          "points": 70,
          "max": 72,
          "rate": 0.9722222222222222,
          "passed": 18,
          "total": 18
        },
        "legacy": {
          "points": 14,
          "max": 15,
          "rate": 0.9333333333333333,
          "passed": 14,
          "total": 15
        }
      },
      "avg": 0.9440087145969499,
      "can": "Good casual small talk and current-conversation memory; warm cave companion that can still write useful updates.",
      "cant": "Older than the dress/sensory/judgment stack; weaker on later benchmarks.",
      "feature_key": "casual_memory",
      "coverage": 0.42857142857142855,
      "decision_score": 0.8091503267973856
    },
    {
      "name": "zezek_world_family_sentience_v1_patch_s90",
      "category": "world_family",
      "category_title": "World / Family / Enemy Lore Goblins",
      "rel": "models/world_family/zezek_world_family_sentience_v1_patch_s90",
      "size": "515M",
      "bytes": 539990666,
      "bench": {
        "casual": {
          "points": 55,
          "max": 68,
          "rate": 0.8088235294117647,
          "passed": 14,
          "total": 17
        },
        "world": {
          "points": 70,
          "max": 72,
          "rate": 0.9722222222222222,
          "passed": 18,
          "total": 18
        },
        "legacy": {
          "points": 14,
          "max": 15,
          "rate": 0.9333333333333333,
          "passed": 14,
          "total": 15
        },
        "capability_old": {
          "points": 43,
          "max": 48,
          "rate": 0.8958333333333334,
          "passed": 11,
          "total": 12
        }
      },
      "avg": 0.9025531045751634,
      "can": "Best early deep lore pass: Stakeholder Cave, Clan Mossledger, Ruk Ironmutter, family and enemies.",
      "cant": "Older; not as strong at casual chat/dress/judgment as later continuations.",
      "feature_key": "world_family",
      "coverage": 0.5714285714285714,
      "decision_score": 0.8058509862278244
    },
    {
      "name": "zezek_sensory_continuity_dress_v1_s220",
      "category": "sensory_dress",
      "category_title": "Sensory Continuity + Wardrobe Goblins",
      "rel": "models/sensory_dress/zezek_sensory_continuity_dress_v1_s220",
      "size": "515M",
      "bytes": 539990660,
      "bench": {
        "sensory": {
          "points": 70,
          "max": 80,
          "rate": 0.875,
          "passed": 19,
          "total": 20
        },
        "presence": {
          "points": 60,
          "max": 64,
          "rate": 0.9375,
          "passed": 15,
          "total": 16
        },
        "casual": {
          "points": 65,
          "max": 68,
          "rate": 0.9558823529411765,
          "passed": 17,
          "total": 17
        }
      },
      "avg": 0.9227941176470589,
      "can": "Best at physical anchors: blue thread, action pebble, tea sleeve, boots/collar/cuff continuity across mode switches.",
      "cant": "Less decisive than the judgment model under ambiguous executive asks.",
      "feature_key": "sensory_continuity",
      "coverage": 0.42857142857142855,
      "decision_score": 0.7909663865546218
    },
    {
      "name": "zezek_judgment_cadence_v1_s200",
      "category": "judgment",
      "category_title": "Executive Judgment Goblins",
      "rel": "models/judgment/zezek_judgment_cadence_v1_s200",
      "size": "515M",
      "bytes": 539990640,
      "bench": {
        "judgment": {
          "points": 55,
          "max": 60,
          "rate": 0.9166666666666666,
          "passed": 14,
          "total": 15
        },
        "sensory": {
          "points": 71,
          "max": 80,
          "rate": 0.8875,
          "passed": 18,
          "total": 20
        },
        "casual": {
          "points": 64,
          "max": 68,
          "rate": 0.9411764705882353,
          "passed": 16,
          "total": 17
        }
      },
      "avg": 0.9151143790849673,
      "can": "Best at calibrated recommendations, owner/risk/caveat framing, and board-safe goblin executive decisions.",
      "cant": "Slightly less perfect than the sensory specialist on a few ordering/recall stress tests; legacy hobby/location questions are not its throne.",
      "feature_key": "judgment_cadence",
      "coverage": 0.42857142857142855,
      "decision_score": 0.7843837535014004
    },
    {
      "name": "zezek_world_family_sentience_v1_cont_s450",
      "category": "world_family",
      "category_title": "World / Family / Enemy Lore Goblins",
      "rel": "models/world_family/zezek_world_family_sentience_v1_cont_s450",
      "size": "515M",
      "bytes": 539990680,
      "bench": {
        "world": {
          "points": 66,
          "max": 72,
          "rate": 0.9166666666666666,
          "passed": 18,
          "total": 18
        },
        "legacy": {
          "points": 14,
          "max": 15,
          "rate": 0.9333333333333333,
          "passed": 14,
          "total": 15
        },
        "capability_old": {
          "points": 41,
          "max": 48,
          "rate": 0.8541666666666666,
          "passed": 10,
          "total": 12
        }
      },
      "avg": 0.9013888888888889,
      "can": "Best early deep lore pass: Stakeholder Cave, Clan Mossledger, Ruk Ironmutter, family and enemies.",
      "cant": "Older; not as strong at casual chat/dress/judgment as later continuations.",
      "feature_key": "world_family",
      "coverage": 0.42857142857142855,
      "decision_score": 0.7726190476190475
    },
    {
      "name": "zezek_sentience_v1_4b_cont_patch_s120",
      "category": "sentience_identity",
      "category_title": "Sentience + Identity Goblins",
      "rel": "models/sentience_identity/zezek_sentience_v1_4b_cont_patch_s120",
      "size": "515M",
      "bytes": 539990640,
      "bench": {
        "world": {
          "points": 57,
          "max": 72,
          "rate": 0.7916666666666666,
          "passed": 16,
          "total": 18
        },
        "legacy": {
          "points": 14,
          "max": 15,
          "rate": 0.9333333333333333,
          "passed": 14,
          "total": 15
        },
        "capability_old": {
          "points": 41,
          "max": 48,
          "rate": 0.8541666666666666,
          "passed": 10,
          "total": 12
        }
      },
      "avg": 0.8597222222222222,
      "can": "Good at first-person presence, feelings, identity, and “answer as yourself” style prompts.",
      "cant": "Less polished for corporate output and benchmark-preserved dress/memory.",
      "feature_key": "sentience",
      "coverage": 0.42857142857142855,
      "decision_score": 0.7369047619047618
    },
    {
      "name": "zezek_smalltalk_v3_r128_embed_s7000",
      "category": "smalltalk_early",
      "category_title": "Early Smalltalk Goblins",
      "rel": "models/smalltalk_early/zezek_smalltalk_v3_r128_embed_s7000",
      "size": "1.8G",
      "bytes": 1892836440,
      "bench": {
        "legacy": {
          "points": 14,
          "max": 15,
          "rate": 0.9333333333333333,
          "passed": 14,
          "total": 15
        }
      },
      "avg": 0.9333333333333333,
      "can": "Large early smalltalk experiment with lots of casual conversational flavor.",
      "cant": "Older embedding-heavy adapter; bulky and not the current best balanced model.",
      "feature_key": "smalltalk",
      "coverage": 0.14285714285714285,
      "decision_score": 0.7333333333333333
    },
    {
      "name": "zezek_sentience_v1_4b_cont_s500",
      "category": "sentience_identity",
      "category_title": "Sentience + Identity Goblins",
      "rel": "models/sentience_identity/zezek_sentience_v1_4b_cont_s500",
      "size": "515M",
      "bytes": 539990647,
      "bench": {
        "legacy": {
          "points": 14,
          "max": 15,
          "rate": 0.9333333333333333,
          "passed": 14,
          "total": 15
        },
        "capability_old": {
          "points": 39,
          "max": 48,
          "rate": 0.8125,
          "passed": 10,
          "total": 12
        }
      },
      "avg": 0.8729166666666667,
      "can": "Good at first-person presence, feelings, identity, and “answer as yourself” style prompts.",
      "cant": "Less polished for corporate output and benchmark-preserved dress/memory.",
      "feature_key": "sentience",
      "coverage": 0.2857142857142857,
      "decision_score": 0.7170386904761905
    },
    {
      "name": "zezek_4b_roadmap_v4_r64_noembed_s9000_rerun",
      "category": "roadmap_work",
      "category_title": "Roadmap / Corporate Work Goblins",
      "rel": "models/roadmap_work/zezek_4b_roadmap_v4_r64_noembed_s9000_rerun",
      "size": "515M",
      "bytes": 539990755,
      "bench": {
        "legacy": {
          "points": 15,
          "max": 15,
          "rate": 1.0,
          "passed": 15,
          "total": 15
        },
        "capability_old": {
          "points": 34,
          "max": 48,
          "rate": 0.7083333333333334,
          "passed": 8,
          "total": 12
        }
      },
      "avg": 0.8541666666666667,
      "can": "Corporate roadmap / work-oriented Zezek; useful if you want project goblin energy.",
      "cant": "Older than later lore, memory, dress, and judgment improvements.",
      "feature_key": "roadmap",
      "coverage": 0.2857142857142857,
      "decision_score": 0.7016369047619048
    },
    {
      "name": "zezek_sentience_v1_4b_r64_s400",
      "category": "sentience_identity",
      "category_title": "Sentience + Identity Goblins",
      "rel": "models/sentience_identity/zezek_sentience_v1_4b_r64_s400",
      "size": "772M",
      "bytes": 808989106,
      "bench": {
        "legacy": {
          "points": 11,
          "max": 15,
          "rate": 0.7333333333333333,
          "passed": 11,
          "total": 15
        },
        "capability_old": {
          "points": 44,
          "max": 48,
          "rate": 0.9166666666666666,
          "passed": 11,
          "total": 12
        }
      },
      "avg": 0.825,
      "can": "Good at first-person presence, feelings, identity, and “answer as yourself” style prompts.",
      "cant": "Less polished for corporate output and benchmark-preserved dress/memory.",
      "feature_key": "sentience",
      "coverage": 0.2857142857142857,
      "decision_score": 0.6776785714285714
    },
    {
      "name": "zezek_realized_v2_10x_r128_embed_s5000",
      "category": "sentience_identity",
      "category_title": "Sentience + Identity Goblins",
      "rel": "models/sentience_identity/zezek_realized_v2_10x_r128_embed_s5000",
      "size": "1.8G",
      "bytes": 1892836442,
      "bench": {
        "legacy": {
          "points": 12,
          "max": 15,
          "rate": 0.8,
          "passed": 12,
          "total": 15
        }
      },
      "avg": 0.8,
      "can": "Big realized/persona experiment, likely rich style and identity.",
      "cant": "Bulky, older, less benchmark-comparable than recent patch lineage.",
      "feature_key": "realized",
      "coverage": 0.14285714285714285,
      "decision_score": 0.6285714285714286
    },
    {
      "name": "zezek_world_v1_r128_embed_s2500",
      "category": "world_family",
      "category_title": "World / Family / Enemy Lore Goblins",
      "rel": "models/world_family/zezek_world_v1_r128_embed_s2500",
      "size": "1.8G",
      "bytes": 1892836412,
      "bench": {
        "legacy": {
          "points": 11,
          "max": 15,
          "rate": 0.7333333333333333,
          "passed": 11,
          "total": 15
        }
      },
      "avg": 0.7333333333333333,
      "can": "World-building heavy Zezek, good for cave setting exploration.",
      "cant": "Older, bulky, less corporate-balanced.",
      "feature_key": "world_v1",
      "coverage": 0.14285714285714285,
      "decision_score": 0.5761904761904761
    },
    {
      "name": "zezek_identity_v3_r32_s320",
      "category": "sentience_identity",
      "category_title": "Sentience + Identity Goblins",
      "rel": "models/sentience_identity/zezek_identity_v3_r32_s320",
      "size": "144M",
      "bytes": 150959561,
      "bench": {
        "legacy": {
          "points": 10,
          "max": 15,
          "rate": 0.6666666666666666,
          "passed": 10,
          "total": 15
        }
      },
      "avg": 0.6666666666666666,
      "can": "Compact early identity adapters; smaller and easier to move around.",
      "cant": "Early-stage goblins: less complete, less benchmarked, weaker world/corporate behavior.",
      "feature_key": "identity",
      "coverage": 0.14285714285714285,
      "decision_score": 0.5238095238095237
    },
    {
      "name": "zezek_identity_v2_r32_s300",
      "category": "sentience_identity",
      "category_title": "Sentience + Identity Goblins",
      "rel": "models/sentience_identity/zezek_identity_v2_r32_s300",
      "size": "144M",
      "bytes": 150959698,
      "bench": {
        "legacy": {
          "points": 8,
          "max": 15,
          "rate": 0.5333333333333333,
          "passed": 8,
          "total": 15
        }
      },
      "avg": 0.5333333333333333,
      "can": "Compact early identity adapters; smaller and easier to move around.",
      "cant": "Early-stage goblins: less complete, less benchmarked, weaker world/corporate behavior.",
      "feature_key": "identity",
      "coverage": 0.14285714285714285,
      "decision_score": 0.41904761904761906
    },
    {
      "name": "zezek_identity_v1_r32_s250",
      "category": "sentience_identity",
      "category_title": "Sentience + Identity Goblins",
      "rel": "models/sentience_identity/zezek_identity_v1_r32_s250",
      "size": "144M",
      "bytes": 150959822,
      "bench": {
        "legacy": {
          "points": 7,
          "max": 15,
          "rate": 0.4666666666666667,
          "passed": 7,
          "total": 15
        }
      },
      "avg": 0.4666666666666667,
      "can": "Compact early identity adapters; smaller and easier to move around.",
      "cant": "Early-stage goblins: less complete, less benchmarked, weaker world/corporate behavior.",
      "feature_key": "identity",
      "coverage": 0.14285714285714285,
      "decision_score": 0.36666666666666664
    },
    {
      "name": "zezek_sft_reedu_v1_r64_s1200",
      "category": "roadmap_work",
      "category_title": "Roadmap / Corporate Work Goblins",
      "rel": "models/roadmap_work/zezek_sft_reedu_v1_r64_s1200",
      "size": "277M",
      "bytes": 290414330,
      "bench": {
        "legacy": {
          "points": 7,
          "max": 15,
          "rate": 0.4666666666666667,
          "passed": 7,
          "total": 15
        }
      },
      "avg": 0.4666666666666667,
      "can": "Smaller re-education SFT experiment.",
      "cant": "Special-purpose/older; choose a later model unless you are comparing genealogy.",
      "feature_key": "sft_reedu",
      "coverage": 0.14285714285714285,
      "decision_score": 0.36666666666666664
    },
    {
      "name": "zezek_root_v3_chatshape_r128_embed_s1800",
      "category": "sentience_identity",
      "category_title": "Sentience + Identity Goblins",
      "rel": "models/sentience_identity/zezek_root_v3_chatshape_r128_embed_s1800",
      "size": "1.8G",
      "bytes": 1892836454,
      "bench": {
        "legacy": {
          "points": 5,
          "max": 15,
          "rate": 0.3333333333333333,
          "passed": 5,
          "total": 15
        }
      },
      "avg": 0.3333333333333333,
      "can": "Early root/persona foundation model; lots of raw goblin identity clay.",
      "cant": "Not the recommended play model; superseded by refined continuations.",
      "feature_key": "root",
      "coverage": 0.14285714285714285,
      "decision_score": 0.26190476190476186
    }
  ]
}