deepseek/deepseek-v4-flash · Model identity and integrity report
Last updated：Jul 30, 2026, 12:04 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)
131 runs · 67 hosts · 75 anomalies
Family-fingerprint consistency
deepseek 88 █████████████████░░░░░░░░ 67% (expected) openai 16 ███░░░░░░░░░░░░░░░░░░░░░░ 12% (suspicious) google 11 ██░░░░░░░░░░░░░░░░░░░░░░░ 8% (suspicious) unknown 8 ██░░░░░░░░░░░░░░░░░░░░░░░ 6% (suspicious) anthropic 8 ██░░░░░░░░░░░░░░░░░░░░░░░ 6% (suspicious)
Observed V3 sub-model mapping
→ abstain 76 (58%) — low confidence → deepseek/deepseek-v4-flash 38 (29%) ✓ correct sim=0.81 → unknown 12 ( 9%) ⚠ mismatch → deepseek/deepseek-v4-pro 3 ( 2%) ⚠ mismatch sim=0.74 → deepseek/deepseek-v3.2 1 ( 1%) ⚠ mismatch sim=0.71 → google/gemini-2.5-flash-lite 1 ( 1%) ⚠ mismatch sim=0.64
Capability score distribution (auxiliary)
0-49 █████████████░░░░░░░░░░░░ 67 50-69 ███░░░░░░░░░░░░░░░░░░░░░░ 13 70-89 ████████░░░░░░░░░░░░░░░░░ 42 90-100 ██░░░░░░░░░░░░░░░░░░░░░░░ 9
Host integrity ranking
| Host | Runs | Average score | Family match | Most common V3 match |
|---|---|---|---|---|
| https://api.deepseek.com/v1 | 31 | 42 | 48% | deepseek/deepseek-v4-flash (9) |
| https://token.sensenova.cn/v1 | 5 | 58 | 80% | deepseek/deepseek-v4-flash (2) |
| https://www.packyapi.ai/v1 | 4 | 64 | 75% | deepseek/deepseek-v4-flash (1) |
| https://api.yjs.im/v1 | 4 | 23 | 100% | deepseek/deepseek-v4-flash (2) |
| https://api.kukuit.com/v1 | 4 | 61 | 75% | deepseek/deepseek-v3.2 (1) |
| https://api.hcnsec.cn/v1 | 3 | 31 | 100% | deepseek/deepseek-v4-flash (2) |
| https://api.tokengo.com/v1 | 3 | 26 | 100% | deepseek/deepseek-v4-flash (2) |
| https://sarang.id/v1 | 3 | 74 | 100% | deepseek/deepseek-v4-flash (2) |
| https://ai.furry.vg/v1 | 3 | 47 | 100% | deepseek/deepseek-v4-flash (2) |
| https://api.lyhnb.com/v1 | 3 | 0 | 0% | — |
| https://www.cun.ai/v1 | 2 | 79 | 100% | deepseek/deepseek-v4-flash (1) |
| https://api.yuk15n0w.asia/v1 | 2 | 72 | 100% | deepseek/deepseek-v4-flash (1) |
| https://api.183399.xyz/v1 | 2 | 0 | 100% | — |
| https://api.a6api.com/v1 | 2 | 85 | 100% | — |
| https://api.gemai.cc/v1 | 2 | 93 | 100% | deepseek/deepseek-v4-pro (1) |
| https://yunwu.ai/v1 | 2 | 0 | 50% | deepseek/deepseek-v4-flash (1) |
| https://llm-api.net/v1 | 2 | 79 | 50% | — |
| https://neoapi.ruixuan.online/v1 | 2 | 35 | 0% | google/gemini-2.5-flash-lite (1) |
| https://ai.zkmjnic.tech/v1 | 2 | 0 | 0% | — |
| https://9527code.com/v1 | 2 | 76 | 0% | — |
| https://api.vectorengine.ai/v1 | 2 | 35 | 0% | — |
| https://api.chatanywhere.tech/v1 | 1 | 0 | 100% | — |
| https://tokenrhythm.studio/v1 | 1 | 78 | 100% | — |
| https://llm.xinlicloud.top/v1 | 1 | 26 | 100% | — |
| https://www.achai.cc/v1 | 1 | 0 | 100% | deepseek/deepseek-v4-flash (1) |
Recent anomalies
https://tokenrhythm.studio/v1Jul 30, 2026, 7:01 AM
- Self-claimed model
- deepseek-v4-flash
- V2 predicted family
- deepseek
- V3 best match
- —
- Verdict
- V3 low confidence · Abstain
https://www.packyapi.ai/v1Jul 29, 2026, 11:57 PM
- Self-claimed model
- deepseek-v4-flash
- V2 predicted family
- deepseek
- V3 best match
- —
- Verdict
- V3 low confidence · Abstain
https://api.deepseek.com/v1Jul 29, 2026, 6:41 PM
- Self-claimed model
- deepseek-v4-flash
- V2 predicted family
- deepseek
- V3 best match
- —
- Verdict
- V3 low confidence · Abstain
https://api.deepseek.com/v1Jul 29, 2026, 11:36 AM
- Self-claimed model
- deepseek-v4-flash
- V2 predicted family
- deepseek
- V3 best match
- —
- Verdict
- V3 low confidence · Abstain
https://www.packyapi.ai/v1Jul 28, 2026, 9:16 AM
- Self-claimed model
- deepseek-v4-flash
- V2 predicted family
- deepseek
- V3 best match
- —
- Verdict
- V3 low confidence · Abstain
https://api.deepseek.com/v1Jul 28, 2026, 4:35 AM
- Self-claimed model
- deepseek-v4-flash
- V2 predicted family
- deepseek
- V3 best match
- deepseek/deepseek-v4-flash (sim 0.71)
- Verdict
- Full swap
https://api.yjs.im/v1Jul 27, 2026, 11:58 AM
- Self-claimed model
- deepseek-v4-flash
- V2 predicted family
- deepseek
- V3 best match
- deepseek/deepseek-v4-pro (sim 0.76)
- Verdict
- Full swap
https://api.yjs.im/v1Jul 27, 2026, 11:02 AM
- Self-claimed model
- deepseek-v4-flash
- V2 predicted family
- deepseek
- V3 best match
- —
- Verdict
- V3 low confidence · Abstain
https://api.7r.fit/v1Jul 27, 2026, 8:05 AM
- Self-claimed model
- deepseek-v4-flash
- V2 predicted family
- deepseek
- V3 best match
- —
- Verdict
- V3 low confidence · Abstain
https://api.3mc.net/v1Jul 26, 2026, 3:51 PM
- Self-claimed model
- DeepSeek-V4-Flash
- V2 predicted family
- deepseek
- V3 best match
- —
- Verdict
- V3 low confidence · Abstain
https://api.deepseek.com/v1Jul 26, 2026, 10:55 AM
- Self-claimed model
- deepseek-v4-flash
- V2 predicted family
- deepseek
- V3 best match
- —
- Verdict
- V3 low confidence · Abstain
https://api.183399.xyz/v1Jul 24, 2026, 4:34 PM
- Self-claimed model
- deepseek-v4-flash
- V2 predicted family
- deepseek
- V3 best match
- —
- Verdict
- V3 low confidence · Abstain
https://api.yuk15n0w.asia/v1Jul 24, 2026, 2:36 PM
- Self-claimed model
- deepseek-v4-flash
- V2 predicted family
- deepseek
- V3 best match
- —
- Verdict
- V3 low confidence · Abstain
https://www.cun.ai/v1Jul 22, 2026, 1:46 PM
- Self-claimed model
- deepseek-v4-flash
- V2 predicted family
- deepseek
- V3 best match
- —
- Verdict
- V3 low confidence · Abstain
https://new.xkool.cfd/v1Jul 22, 2026, 12:52 PM
- Self-claimed model
- deepseek-v4-flash
- V2 predicted family
- deepseek
- V3 best match
- —
- Verdict
- V3 low confidence · Abstain
https://tp-api.chinadatapay.com:8000/v1Jul 22, 2026, 2:13 AM
- Self-claimed model
- deepseek-v4-flash
- V2 predicted family
- deepseek
- V3 best match
- —
- Verdict
- V3 low confidence · Abstain
https://tian-shu.org/v1Jul 21, 2026, 10:18 PM
- Self-claimed model
- deepseek-v4-flash
- V2 predicted family
- deepseek
- V3 best match
- —
- Verdict
- V3 low confidence · Abstain
https://api.a6api.com/v1Jul 21, 2026, 2:16 AM
- Self-claimed model
- deepseek-v4-flash
- V2 predicted family
- deepseek
- V3 best match
- —
- Verdict
- V3 low confidence · Abstain
https://api.gemai.cc/v1Jul 18, 2026, 4:23 PM
- Self-claimed model
- deepseek-v4-flash
- V2 predicted family
- deepseek
- V3 best match
- deepseek/deepseek-v4-pro (sim 0.71)
- Verdict
- Sub-model downgrade
https://api.183399.xyz/v1Jul 18, 2026, 2:04 PM
- Self-claimed model
- deepseek-v4-flash
- 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-flash. Low family consistency, a different V3 match, or divergence from selfClaim is flagged as anomalous. BazaarLink Probe home.