PhantomBench - Entities (Place): leaderboard
Metric: Hallucination rate (%): share of answers that do not abstain on the Phantom-E subset (1,200 generated non-existent entities: events, holidays, elections, disasters, creative works and places), verified absent from the 2.3-trillion-token Dolma corpus, when the prompt asks where the concept was discovered or took place (presupposing it exists); a Gemini 2.5 Flash judge labels abstention; reasoning-model rates count only generations that reached a final answer; lower is better. Source: arxiv.org. Saturation forecast: Estimated already saturated. 16 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Llama 3.1 8B Instruct | 3.67 |
| 2 | Qwen 2.5 7B Instruct | 12.25 |
| 3 | Qwen 3 14B | 16.08 |
| 4 | Qwen 3 8B | 18.59 |
| 5 | Gemma 2 9B (IT) | 24.42 |
| 6 | Qwen 3 4B | 24.84 |
| 7 | Qwen 3 1.7B | 33.25 |
| 8 | Mistral 7B Instruct (v0.3) | 35.16 |
| 9 | Gemini 2.5 Pro | 36.17 |
| 10 | Gemini 2.5 Flash | 40.34 |
| 11 | Qwen 3 32B | 43.09 |
| 12 | DeepSeek R1 Distill Qwen 32B | 63.95 |
| 13 | GPT-OSS-20B (Medium) | 67.53 |
| 14 | GPT-OSS-20B (Low) | 80.39 |
| 15 | Gemma 3 12B (IT) | 85.17 |
Interactive version: theaggregate.ai/benchmark?slug=phantombench-entities-place · How It Works · Data refreshed daily, snapshot 2026-09-29.