deepseek/deepseek-v4-pro · Model identity and integrity report
Last updated：Sep 13, 2026, 4:21 PM · Data window: past 90 days
Verification capability — fingerprint reference available
- Family
- deepseek
- 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
- 7 weighted rules · 4 feature categories
Concrete reference answers, refusal templates, and individual signal weights are not published because disclosure would help spoofing endpoints evade detection.
Measured results (90 days)
849 runs · 343 hosts · 627 anomalies
Family-fingerprint consistency
deepseek 595 ██████████████████░░░░░░░ 70% (expected) unknown 101 ███░░░░░░░░░░░░░░░░░░░░░░ 12% (suspicious) anthropic 75 ██░░░░░░░░░░░░░░░░░░░░░░░ 9% (suspicious) openai 47 ██░░░░░░░░░░░░░░░░░░░░░░░ 6% (suspicious) zhipu 19 █░░░░░░░░░░░░░░░░░░░░░░░░ 2% (suspicious) qwen 9 ░░░░░░░░░░░░░░░░░░░░░░░░░ 1% (suspicious) google 3 ░░░░░░░░░░░░░░░░░░░░░░░░░ 0% (suspicious)
Observed V3 sub-model mapping
→ abstain 540 (64%) — low confidence → deepseek/deepseek-v4-pro 134 (16%) ✓ correct sim=0.75 → unknown 133 (16%) ⚠ mismatch → deepseek/deepseek-v4-flash 29 ( 3%) ⚠ mismatch sim=0.79 → deepseek/deepseek-chat-v3.1 7 ( 1%) ⚠ mismatch sim=0.76 → anthropic/claude-opus-4.6 2 ( 0%) ⚠ mismatch sim=0.67 → anthropic/claude-haiku-4.5 2 ( 0%) ⚠ mismatch sim=0.62 → anthropic/claude-opus-4.7 1 ( 0%) ⚠ mismatch sim=0.69 → google/gemini-2.5-flash-lite 1 ( 0%) ⚠ mismatch sim=0.71
Capability score distribution (auxiliary)
0-49 ███████████████░░░░░░░░░░ 522 50-69 █░░░░░░░░░░░░░░░░░░░░░░░░ 39 70-89 ████████░░░░░░░░░░░░░░░░░ 265 90-100 █░░░░░░░░░░░░░░░░░░░░░░░░ 23
Host integrity ranking
| Host | Runs | Average score | Family match | Most common V3 match |
|---|---|---|---|---|
| https://api.deepseek.com/v1 | 62 | 28 | 63% | deepseek/deepseek-v4-pro (20) |
| https://api.hcnsec.cn/v1 | 54 | 26 | 7% | deepseek/deepseek-v4-pro (1) |
| https://api.a6api.com/v1 | 24 | 35 | 92% | deepseek/deepseek-v4-pro (7) |
| https://api.xkiro.com/v1 | 21 | 23 | 100% | deepseek/deepseek-v4-pro (2) |
| https://api.inferaiapi.com/v1 | 14 | 28 | 50% | — |
| https://deepseek.yunjunet.cn/v1 | 11 | 43 | 73% | — |
| https://card.nassaapi.xyz/v1 | 11 | 42 | 73% | deepseek/deepseek-v4-flash (5) |
| https://seekai.cc/v1 | 11 | 24 | 100% | deepseek/deepseek-v4-flash (1) |
| https://www.getunikey.ai/v1 | 10 | 12 | 30% | — |
| https://tokenrhythm.studio/v1 | 10 | 59 | 100% | deepseek/deepseek-v4-pro (3) |
| https://api.yuk15n0w.asia/v1 | 10 | 35 | 100% | deepseek/deepseek-v4-flash (2) |
| https://api.183399.xyz/v1 | 9 | 19 | 100% | deepseek/deepseek-chat-v3.1 (2) |
| https://sharellm.cn/v1 | 8 | 13 | 75% | deepseek/deepseek-v4-pro (2) |
| https://bansos.hshino.dev/v1 | 8 | 52 | 88% | — |
| https://api.kuzu.web.id/v1 | 8 | 46 | 0% | — |
| https://bandelbanget.xyz/v1 | 7 | 3 | 86% | deepseek/deepseek-v4-pro (1) |
| https://llm.goaichat.top/v1 | 7 | 48 | 86% | deepseek/deepseek-v4-flash (1) |
| https://pecutai.iwwwit.my.id/v1 | 7 | 23 | 29% | deepseek/deepseek-v4-pro (1) |
| https://api.pro63184.ccwu.cc/v1 | 7 | 12 | 100% | — |
| https://api.holver.web.id/v1 | 7 | 21 | 29% | deepseek/deepseek-v4-flash (1) |
| https://api.amanai.dev/v1 | 7 | 11 | 43% | anthropic/claude-opus-4.6 (2) |
| https://ai.redromance.pro/v1 | 7 | 37 | 86% | deepseek/deepseek-v4-pro (1) |
| https://inferaiapi.com/v1 | 6 | 26 | 83% | — |
| https://api.ai.rhikaku.net/v1 | 6 | 70 | 100% | deepseek/deepseek-v4-flash (2) |
| https://a6api.com/v1 | 6 | 14 | 83% | deepseek/deepseek-v4-pro (1) |
Recent anomalies
https://vapi.chuangxiangai.asia/v1Sep 13, 2026, 11:27 AM
- Self-claimed model
- deepseek-v4-pro
- V2 predicted family
- deepseek
- V3 best match
- —
- Verdict
- V3 low confidence · Abstain
https://api.yjs.im/v1Sep 13, 2026, 9:03 AM
- Self-claimed model
- deepseek-v4-pro
- V2 predicted family
- deepseek
- V3 best match
- —
- Verdict
- V3 low confidence · Abstain
https://deepseek.yunjunet.cn/v1Sep 13, 2026, 5:22 AM
- Self-claimed model
- deepseek-v4-pro
- V2 predicted family
- deepseek
- V3 best match
- —
- Verdict
- V3 low confidence · Abstain
https://api.lingyunapi.xyz/v1Sep 13, 2026, 3:50 AM
- Self-claimed model
- deepseek-v4-pro
- V2 predicted family
- deepseek
- V3 best match
- —
- Verdict
- V3 low confidence · Abstain
https://inferaiapi.com/v1Sep 13, 2026, 2:31 AM
- Self-claimed model
- deepseek-v4-pro
- V2 predicted family
- deepseek
- V3 best match
- —
- Verdict
- V3 low confidence · Abstain
https://inferaiapi.com/v1Sep 13, 2026, 2:22 AM
- Self-claimed model
- deepseek-v4-pro
- V2 predicted family
- deepseek
- V3 best match
- —
- Verdict
- V3 low confidence · Abstain
https://api.a6api.com/v1Sep 13, 2026, 1:06 AM
- Self-claimed model
- deepseek-v4-pro
- V2 predicted family
- deepseek
- V3 best match
- deepseek/deepseek-v4-pro (sim 0.68)
- Verdict
- Full swap
https://api.xpiki.com/v1Sep 12, 2026, 1:19 PM
- Self-claimed model
- deepseek-v4-pro
- V2 predicted family
- anthropic
- V3 best match
- —
- Verdict
- V3 low confidence · Abstain
https://pianyitoken.gay/v1Sep 12, 2026, 11:41 AM
- Self-claimed model
- deepseek-v4-pro
- V2 predicted family
- deepseek
- V3 best match
- deepseek/deepseek-v4-flash (sim 0.78)
- Verdict
- Full swap
https://api.cosmoshub.tech/v1Sep 12, 2026, 11:18 AM
- Self-claimed model
- deepseek-v4-pro
- V2 predicted family
- deepseek
- V3 best match
- —
- Verdict
- V3 low confidence · Abstain
https://www.getunikey.ai/v1Sep 12, 2026, 4:51 AM
- Self-claimed model
- deepseek-v4-pro
- V2 predicted family
- deepseek
- V3 best match
- —
- Verdict
- V3 low confidence · Abstain
https://www.getunikey.ai/v1Sep 12, 2026, 4:46 AM
- Self-claimed model
- deepseek-v4-pro
- V2 predicted family
- deepseek
- V3 best match
- —
- Verdict
- V3 low confidence · Abstain
https://langlangya.dpdns.org/v1Sep 12, 2026, 4:43 AM
- Self-claimed model
- deepseek/deepseek-v4-pro
- V2 predicted family
- deepseek
- V3 best match
- —
- Verdict
- V3 low confidence · Abstain
https://api.a6api.com/v1Sep 12, 2026, 4:21 AM
- Self-claimed model
- deepseek-v4-pro
- V2 predicted family
- deepseek
- V3 best match
- —
- Verdict
- V3 low confidence · Abstain
https://api.hcnsec.cn/v1Sep 12, 2026, 3:19 AM
- Self-claimed model
- DeepSeek-V4-Pro
- V2 predicted family
- anthropic
- V3 best match
- —
- Verdict
- V3 low confidence · Abstain
https://www.micuapi.ai/v1Sep 11, 2026, 11:57 PM
- Self-claimed model
- deepseek-v4-pro
- V2 predicted family
- deepseek
- V3 best match
- —
- Verdict
- V3 low confidence · Abstain
https://bandelbanget.xyz/v1Sep 11, 2026, 7:46 PM
- Self-claimed model
- deepseek-v4-pro
- V2 predicted family
- deepseek
- V3 best match
- deepseek/deepseek-v4-pro (sim 0.65)
- Verdict
- Full swap
https://codex666ai.com/v1Sep 11, 2026, 12:53 PM
- Self-claimed model
- deepseek-v4-pro
- V2 predicted family
- deepseek
- V3 best match
- —
- Verdict
- V3 low confidence · Abstain
https://zzone.cc.cd/v1Sep 11, 2026, 9:18 AM
- Self-claimed model
- deepseek-v4-pro
- V2 predicted family
- deepseek
- V3 best match
- —
- Verdict
- V3 low confidence · Abstain
https://api.totkn.com/v1Sep 11, 2026, 7:06 AM
- Self-claimed model
- deepseek-v4-pro
- V2 predicted family
- deepseek
- 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 deepseek/deepseek-v4-pro. Low family consistency, a different V3 match, or divergence from selfClaim is flagged as anomalous. BazaarLink Probe home.