Skip to content
Panshi
EN / /
← Panshi Sentinel
Panshi Sentinel · point-in-time verification

CCSub

Not certified · 1 of 3 mainstream models verified, 2 did not pass

www.ccsub.net · checked 2026-06-22 · engine v0.8

⚠️ 存在未通过验真的模型(跨厂不符) → 封顶C, 不展示认证微章

Provider info

Name
CCSub
API endpoint
https://www.ccsub.net/v1
Last checked
2026-06-22
34.4
trust index
Not certified verify rate 1/3
Authenticity34.4
Crypto / cloud confirm0
Coverage100
Availability100

Per-model verdict

Each model is judged independently — one verified model never speaks for the whole relay. Models outside our reference set are marked "not yet covered" with no verdict.

gpt-5.5
detected: openai
75
Genuine · multi-dimensional confirmation
Behavioral fingerprint3 cross-checked signals
Multi-signal analysis consistent, no contradiction 2089ms
claude-opus-4-8
detected: deepseek
10
Failed verification (behaves like a different model)
Behavioral fingerprint
Multi-signal analysis: upstream inconsistent with claimed model 2228ms
claude-sonnet-4-6
detected: deepseek
10
Failed verification (behaves like a different model)
Behavioral fingerprint
Multi-signal analysis: upstream inconsistent with claimed model 2049ms
How to read this grade

🛡️ A · Panshi Certified — mainstream models verified with strong evidence (trust ≥ 90), no substitution or downgrade.

B · Verified — mainstream models are genuine (trust ≥ 75), no substitution or downgrade.

🟡 C · Partial — some models unconfirmed / downgrade suspected.

Not certified — at least one mainstream model did not pass verification. Capping rule: any cross-vendor mismatch caps the whole relay to "not certified"; any same-vendor downgrade caps it to B — so a few genuine models can never mask a substituted one.

How we test

1

Behavioral fingerprinting

A set of probes samples the model's answer-style distribution and compares it against our official-source reference fingerprints — identifying which model it behaves like, even when prompted to disguise itself as Claude.

2

Cryptographic-grade verification

Supported models undergo cryptographic-grade authenticity verification — the strongest evidence; genuine models resold through Bedrock / Vertex official cloud are cross-confirmed via channel fingerprint and multiple signals.

3

Multi-signal cross-check + per-model verdict

High confidence requires several independent signals (identity, latency, capability, rank tests) to agree. Each model is judged independently — one verified model never speaks for the whole relay.

⚠️ Results are probabilistic signals, not legal proof. This certification is a point-in-time snapshot; a relay's backend may change at any time, and continuous assurance requires paid monitoring. Models outside our reference set are marked "not yet covered" with no verdict. We only publish positive verification — those that do not pass are simply "not certified".

Full scoring method & grades →

Data looks wrong, or want your relay certified? Contact us.