anthropic/claude-haiku-5.5 · Model identity and integrity report

Family

anthropic

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

60 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.

12 runs · 9 hosts · 10 anomalies

Family-fingerprint consistency

  anthropic       10  █████████████████████░░░░   83% (expected)
  unknown          2  ████░░░░░░░░░░░░░░░░░░░░░   17% (suspicious)

Observed V3 sub-model mapping

  → abstain                                  6 (50%) — low confidence
  → anthropic/claude-opus-4.8                4 (33%) ⚠ mismatch sim=0.83
  → unknown                                  2 (17%) ⚠ mismatch

Capability score distribution (auxiliary)

  0-49     █████████████████████████   12
  50-69    ░░░░░░░░░░░░░░░░░░░░░░░░░    0
  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 anthropic/claude-haiku-5.5. Low family consistency, a different V3 match, or divergence from selfClaim is flagged as anomalous. BazaarLink Probe home.