z-ai/glm-5.1 · Model identity and integrity report

Family

zhipu

V3 sub-model reference

Available ✓

Checks

知識截止 · 能力探針 · 拒絕模板 · 拒絕溫度參數

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

118 runs · 67 hosts · 107 anomalies

Family-fingerprint consistency

  zhipu           36  ████████░░░░░░░░░░░░░░░░░   31% (suspicious)
  openai          33  ███████░░░░░░░░░░░░░░░░░░   28% (suspicious)
  anthropic       24  █████░░░░░░░░░░░░░░░░░░░░   20% (suspicious)
  google          13  ███░░░░░░░░░░░░░░░░░░░░░░   11% (suspicious)
  unknown         11  ██░░░░░░░░░░░░░░░░░░░░░░░    9% (suspicious)
  deepseek         1  ░░░░░░░░░░░░░░░░░░░░░░░░░    1% (suspicious)

Observed V3 sub-model mapping

  → abstain                                 72 (61%) — low confidence
  → unknown                                 17 (14%) ⚠ mismatch
  → z-ai/glm-5.1                             7 ( 6%) ✓ correct sim=0.80
  → z-ai/glm-5.2                             7 ( 6%) ⚠ mismatch sim=0.82
  → z-ai/glm-5                               6 ( 5%) ⚠ mismatch sim=0.79
  → google/gemini-3.1-pro-preview            5 ( 4%) ⚠ mismatch sim=0.83
  → deepseek/deepseek-v4-pro                 1 ( 1%) ⚠ mismatch sim=0.78
  → google/gemini-3.1-flash-lite-preview     1 ( 1%) ⚠ mismatch sim=0.73
  → anthropic/claude-opus-4.7                1 ( 1%) ⚠ mismatch sim=0.69
  → anthropic/claude-opus-4.6                1 ( 1%) ⚠ mismatch sim=0.68

Capability score distribution (auxiliary)

  0-49     █████████████░░░░░░░░░░░░   59
  50-69    ██░░░░░░░░░░░░░░░░░░░░░░░    9
  70-89    ██████████░░░░░░░░░░░░░░░   46
  90-100   █░░░░░░░░░░░░░░░░░░░░░░░░    4

BazaarLink Probe combines a Bayesian V2 family classifier across nine fingerprint categories with V3 sub-model references to verify endpoints claiming z-ai/glm-5.1. Low family consistency, a different V3 match, or divergence from selfClaim is flagged as anomalous. BazaarLink Probe home.