k3 · Model identity and integrity report
Last updated：Sep 17, 2026, 6:24 PM · Data window: past 90 days
Measured results (90 days)
13 runs · 9 hosts · 9 anomalies
Family-fingerprint consistency
anthropic 8 ████████████████░░░░░░░░░ 62% (suspicious) unknown 4 ████████░░░░░░░░░░░░░░░░░ 31% (suspicious) openai 1 ██░░░░░░░░░░░░░░░░░░░░░░░ 8% (suspicious)
Observed V3 sub-model mapping
→ unknown 5 (38%) ⚠ mismatch → abstain 5 (38%) — low confidence → anthropic/claude-sonnet-5 2 (15%) ⚠ mismatch sim=0.74 → anthropic/claude-fable-5 1 ( 8%) ⚠ mismatch sim=0.90
Capability score distribution (auxiliary)
0-49 ████████████████░░░░░░░░░ 8 50-69 ██░░░░░░░░░░░░░░░░░░░░░░░ 1 70-89 ████████░░░░░░░░░░░░░░░░░ 4 90-100 ░░░░░░░░░░░░░░░░░░░░░░░░░ 0
Host integrity ranking
| Host | Runs | Average score | Family match | Most common V3 match |
|---|---|---|---|---|
| https://api.kimi.com/coding/v1 | 3 | 0 | 0% | — |
| https://ai-cf.boke.page/v1 | 2 | 79 | 0% | — |
| http://c2846.top:8081/v1 | 2 | 0 | 0% | anthropic/claude-sonnet-5 (1) |
| https://codex666ai.com/v1 | 1 | 0 | 0% | — |
| https://sub.laogou8.com/v1 | 1 | 76 | 0% | — |
| https://ai.linjzy.com/v1 | 1 | 79 | 0% | — |
| https://liliapi.siphot.com/v1 | 1 | 11 | 0% | — |
| https://hemasir.online/v1 | 1 | 68 | 0% | anthropic/claude-sonnet-5 (1) |
| https://sub.sailapi.top/v1 | 1 | 0 | 0% | — |
Recent anomalies
https://codex666ai.com/v1Sep 11, 2026, 12:54 PM
- Self-claimed model
- k3
- V2 predicted family
- openai
- V3 best match
- —
- Verdict
- V3 low confidence · Abstain
https://sub.laogou8.com/v1Sep 3, 2026, 3:04 AM
- Self-claimed model
- k3
- V2 predicted family
- anthropic
- V3 best match
- —
- Verdict
- Family mismatch
https://api.kimi.com/coding/v1Aug 31, 2026, 5:12 PM
- Self-claimed model
- k3
- V2 predicted family
- anthropic
- V3 best match
- —
- Verdict
- V3 low confidence · Abstain
https://ai-cf.boke.page/v1Aug 28, 2026, 3:26 PM
- Self-claimed model
- k3
- V2 predicted family
- anthropic
- V3 best match
- —
- Verdict
- V3 low confidence · Abstain
https://ai-cf.boke.page/v1Aug 26, 2026, 5:57 AM
- Self-claimed model
- k3
- V2 predicted family
- anthropic
- V3 best match
- —
- Verdict
- V3 low confidence · Abstain
https://ai.linjzy.com/v1Aug 25, 2026, 5:29 AM
- Self-claimed model
- k3
- V2 predicted family
- anthropic
- V3 best match
- —
- Verdict
- V3 low confidence · Abstain
https://hemasir.online/v1Jul 27, 2026, 9:05 AM
- Self-claimed model
- k3
- V2 predicted family
- anthropic
- V3 best match
- anthropic/claude-sonnet-5 (sim 0.76)
- Verdict
- Full swap
http://c2846.top:8081/v1Jul 26, 2026, 9:55 AM
- Self-claimed model
- k3
- V2 predicted family
- anthropic
- V3 best match
- anthropic/claude-sonnet-5 (sim 0.72)
- Verdict
- Full swap
http://c2846.top:8081/v1Jul 21, 2026, 6:32 AM
- Self-claimed model
- k3
- V2 predicted family
- anthropic
- V3 best match
- anthropic/claude-fable-5 (sim 0.90)
- Verdict
- Full swap
Methodology
BazaarLink Probe combines a Bayesian V2 family classifier across nine fingerprint categories with V3 sub-model references to verify endpoints claiming k3. Low family consistency, a different V3 match, or divergence from selfClaim is flagged as anomalous. BazaarLink Probe home.