vectara/hallucination-leaderboard

Leaderboard Comparing LLM Performance at Producing Hallucinations when Summarizing Short Documents

62
/ 100
Established

Leverages Vectara's HHEM-2.3 hallucination evaluation model to score LLMs across a curated, non-public dataset of 7700+ documents spanning news, science, medicine, and legal domains with varying complexity (50–24K words). Rankings track hallucination rate, factual consistency, answer rate, and summary length, with results updated regularly as models evolve. Integrates with Hugging Face for interactive exploration and provides an open-source HHEM-2.1 variant for reproducible research.

3,122 stars. Actively maintained with 9 commits in the last 30 days.

No Package No Dependents
Maintenance 20 / 25
Adoption 10 / 25
Maturity 16 / 25
Community 16 / 25

How are scores calculated?

Stars

3,122

Forks

96

Language

Python

License

Apache-2.0

Last pushed

Mar 10, 2026

Commits (30d)

9

Get this data via API

curl "https://pt-edge.onrender.com/api/v1/quality/llm-tools/vectara/hallucination-leaderboard"

Open to everyone — 100 requests/day, no key needed. Get a free key for 1,000/day.