google/gemini-3-flash-preview · Model identity and integrity report
Last updated：Sep 13, 2026, 5:00 PM · Data window: past 90 days
Measured results (90 days)
20 runs · 11 hosts · 14 anomalies
Family-fingerprint consistency
google 12 ███████████████░░░░░░░░░░ 60% (expected) unknown 5 ██████░░░░░░░░░░░░░░░░░░░ 25% (suspicious) openai 2 ███░░░░░░░░░░░░░░░░░░░░░░ 10% (suspicious) anthropic 1 █░░░░░░░░░░░░░░░░░░░░░░░░ 5% (suspicious)
Observed V3 sub-model mapping
→ abstain 9 (45%) — low confidence → unknown 6 (30%) ⚠ mismatch → google/gemini-3.1-pro-preview 3 (15%) ⚠ mismatch sim=0.75 → google/gemini-3.6-flash 2 (10%) ⚠ mismatch sim=0.72
Capability score distribution (auxiliary)
0-49 ███████████████████████░░ 18 50-69 ███░░░░░░░░░░░░░░░░░░░░░░ 2 70-89 ░░░░░░░░░░░░░░░░░░░░░░░░░ 0 90-100 ░░░░░░░░░░░░░░░░░░░░░░░░░ 0
Host integrity ranking
| Host | Runs | Average score | Family match | Most common V3 match |
|---|---|---|---|---|
| https://agonizing-reflex-darkened.ngrok-free.dev/v1 | 4 | 0 | 0% | — |
| https://bboluo.com/v1 | 2 | 0 | 50% | — |
| https://api.gemai.cc/v1 | 2 | 0 | 100% | google/gemini-3.1-pro-preview (1) |
| https://www.anyfast.ai/v1 | 2 | 64 | 100% | google/gemini-3.1-pro-preview (1) |
| https://a6api.com/v1 | 2 | 0 | 100% | — |
| https://api.tokeneasy.com.cn/v1 | 2 | 0 | 50% | — |
| https://seesaw2ai.com/v1 | 2 | 0 | 50% | google/gemini-3.1-pro-preview (1) |
| https://api.apimart.ai/v1 | 1 | 0 | 100% | google/gemini-3.6-flash (1) |
| https://pro.gemai.cc/v1 | 1 | 0 | 100% | google/gemini-3.6-flash (1) |
| https://tian-shu.org/v1 | 1 | 0 | 100% | — |
| https://ai.pianyiapi.com/v1 | 1 | 0 | 0% | — |
Recent anomalies
https://bboluo.com/v1Sep 11, 2026, 12:38 AM
- Self-claimed model
- [L]gemini-3-flash-preview
- V2 predicted family
- V3 best match
- —
- Verdict
- Full swap
https://api.apimart.ai/v1Sep 6, 2026, 2:56 PM
- Self-claimed model
- gemini-3-flash-preview
- V2 predicted family
- V3 best match
- google/gemini-3.6-flash (sim 0.73)
- Verdict
- Full swap
https://pro.gemai.cc/v1Sep 6, 2026, 2:00 PM
- Self-claimed model
- gemini-3-flash-preview
- V2 predicted family
- V3 best match
- google/gemini-3.6-flash (sim 0.73)
- Verdict
- Full swap
https://bboluo.com/v1Aug 11, 2026, 12:45 AM
- Self-claimed model
- (反重力)gemini-3-flash-preview
- V2 predicted family
- openai
- V3 best match
- —
- Verdict
- V3 low confidence · Abstain
https://api.gemai.cc/v1Aug 7, 2026, 10:26 PM
- Self-claimed model
- [福利]gemini-3-flash-preview
- V2 predicted family
- V3 best match
- google/gemini-3.1-pro-preview (sim 0.76)
- Verdict
- Sub-model downgrade
https://www.anyfast.ai/v1Aug 4, 2026, 8:00 AM
- Self-claimed model
- gemini-3-flash-preview
- V2 predicted family
- V3 best match
- google/gemini-3.1-pro-preview (sim 0.76)
- Verdict
- Sub-model downgrade
https://www.anyfast.ai/v1Aug 4, 2026, 3:20 AM
- Self-claimed model
- gemini-3-flash-preview
- V2 predicted family
- V3 best match
- —
- Verdict
- V3 low confidence · Abstain
https://tian-shu.org/v1Aug 2, 2026, 5:59 PM
- Self-claimed model
- gemini-3-flash-preview
- V2 predicted family
- V3 best match
- —
- Verdict
- V3 low confidence · Abstain
https://a6api.com/v1Aug 2, 2026, 5:27 AM
- Self-claimed model
- gemini-3-flash-preview
- V2 predicted family
- V3 best match
- —
- Verdict
- V3 low confidence · Abstain
https://a6api.com/v1Aug 2, 2026, 5:24 AM
- Self-claimed model
- gemini-3-flash-preview
- V2 predicted family
- V3 best match
- —
- Verdict
- V3 low confidence · Abstain
https://api.tokeneasy.com.cn/v1Jul 28, 2026, 6:20 AM
- Self-claimed model
- gemini-3-flash-preview
- V2 predicted family
- V3 best match
- —
- Verdict
- V3 low confidence · Abstain
https://api.tokeneasy.com.cn/v1Jul 28, 2026, 6:18 AM
- Self-claimed model
- gemini-3-flash-preview
- V2 predicted family
- anthropic
- V3 best match
- —
- Verdict
- V3 low confidence · Abstain
https://seesaw2ai.com/v1Jun 28, 2026, 8:39 AM
- Self-claimed model
- gemini-3-flash-preview
- V2 predicted family
- V3 best match
- google/gemini-3.1-pro-preview (sim 0.74)
- Verdict
- Sub-model downgrade
https://ai.pianyiapi.com/v1Jun 24, 2026, 2:20 PM
- Self-claimed model
- gemini-3-flash-preview
- V2 predicted family
- openai
- V3 best match
- —
- Verdict
- V3 low confidence · Abstain
Methodology
BazaarLink Probe combines a Bayesian V2 family classifier across nine fingerprint categories with V3 sub-model references to verify endpoints claiming google/gemini-3-flash-preview. Low family consistency, a different V3 match, or divergence from selfClaim is flagged as anomalous. BazaarLink Probe home.