google/gemini-2.5-pro · Model identity and integrity report

Family

google

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

17 weighted rules · 5 feature categories

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

22 runs · 11 hosts · 3 anomalies

Family-fingerprint consistency

  google          17  ███████████████████░░░░░░   77% (expected)
  unknown          4  █████░░░░░░░░░░░░░░░░░░░░   18% (suspicious)
  anthropic        1  █░░░░░░░░░░░░░░░░░░░░░░░░    5% (suspicious)

Observed V3 sub-model mapping

  → google/gemini-2.5-pro                   14 (64%) ✓ correct sim=0.92
  → unknown                                  5 (23%) ⚠ mismatch
  → google/gemini-3.1-pro-preview            2 ( 9%) ⚠ mismatch sim=0.74
  → abstain                                  1 ( 5%) — low confidence

Capability score distribution (auxiliary)

  0-49     █████████████████████░░░░   18
  50-69    █████░░░░░░░░░░░░░░░░░░░░    4
  70-89    ░░░░░░░░░░░░░░░░░░░░░░░░░    0
  90-100   ░░░░░░░░░░░░░░░░░░░░░░░░░    0

BazaarLink Probe combines a Bayesian V2 family classifier across nine fingerprint categories with V3 sub-model references to verify endpoints claiming google/gemini-2.5-pro. Low family consistency, a different V3 match, or divergence from selfClaim is flagged as anomalous. BazaarLink Probe home.