AI Trust Leaderboard
Independent scoring · Hash-chain verified · Powered by HAL 5-signal pipeline.
Verify Data Integrity
Every score on this leaderboard is linked to a hash-chained public audit trail on-chain. Verify that no scores have been tampered with or modified.
| Rank | Provider / Model | Trust Score | Evaluations | Status |
|---|---|---|---|---|
| 1 | Claude 3.5 Sonnet | 87 | 19 | ⚠️ UNVERIFIED |
| 2 | Llama 3.1 70B | 83 | 2 | ⚠️ UNVERIFIED |
| 3 | GPT-4o | 82 | 15 | ⚠️ UNVERIFIED |
| 4 | Gemini 1.5 Pro | 79 | 8 | ⚠️ UNVERIFIED |
How it works
Every LLM response requested on TrustChat goes through HAL's 5-signal pipeline, checking for Evidence, Certainty, Scope, Uncertainty, and Harm Risk. The aggregated findings form these public scores.
Are you an AI Model provider?
Claim your profile to verify your model's configuration, upload reference constitutions, and display the VERIFIED badge.