gpt54mini · Model identity and integrity report

Family

openai

V3 sub-model reference

Available ✓

Checks

知識截止 · 能力探針 · 拒絕模板 · 拒絕溫度參數

Family-fingerprint coverage

52 weighted rules · 8 feature categories

Concrete reference answers, refusal templates, and individual signal weights are not published because disclosure would help spoofing endpoints evade detection.

107 runs · 63 hosts · 89 anomalies

Family-fingerprint consistency

  openai          85  ████████████████████░░░░░   79% (expected)
  unknown         13  ███░░░░░░░░░░░░░░░░░░░░░░   12% (suspicious)
  anthropic        9  ██░░░░░░░░░░░░░░░░░░░░░░░    8% (suspicious)

Observed V3 sub-model mapping

  → abstain                                 71 (66%) — low confidence
  → openai/gpt-5.4                          16 (15%) ⚠ mismatch sim=0.90
  → unknown                                 15 (14%) ⚠ mismatch
  → openai/gpt-5.4-mini                      3 ( 3%) ⚠ mismatch sim=0.86
  → openai/gpt-4.1-mini                      1 ( 1%) ⚠ mismatch sim=0.70
  → openai/gpt-5.5                           1 ( 1%) ⚠ mismatch sim=0.94

Capability score distribution (auxiliary)

  0-49     █████████████████░░░░░░░░   72
  50-69    ░░░░░░░░░░░░░░░░░░░░░░░░░    1
  70-89    ██████░░░░░░░░░░░░░░░░░░░   26
  90-100   ██░░░░░░░░░░░░░░░░░░░░░░░    8

BazaarLink Probe combines a Bayesian V2 family classifier across nine fingerprint categories with V3 sub-model references to verify endpoints claiming gpt54mini. Low family consistency, a different V3 match, or divergence from selfClaim is flagged as anomalous. BazaarLink Probe home.