LLM Hallucination Benchmark v2: 25K rows, 6 Models
Subtle hallucinations across GPT-4o, Claude, Gemini, Llama3, Mistral & Gemma
What you can do with it
Check whether model answers contain unsupported or incorrect claims.
Download size: 2.4 MB
Resource information
- Data purpose
- Evaluation set
- Task
- Language models and prompts
- Modality
- Text
- Version
- 1
- Updated by publisher
- Apr 30, 2026
- License
- CC BY-SA 4.0
- Attribution
- Original dataset