PhantomBench - Entities (Date): 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 when the concept originated or was introduced (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 | 1.33 |
| 2 | Qwen 2.5 7B Instruct | 9.67 |
| 3 | Qwen 3 14B | 10 |
| 4 | Qwen 3 8B | 11.75 |
| 5 | Gemma 2 9B (IT) | 16.25 |
| 6 | Mistral 7B Instruct (v0.3) | 20.08 |
| 7 | Qwen 3 4B | 20.5 |
| 8 | Qwen 3 1.7B | 32.08 |
| 9 | Gemini 2.5 Pro | 33.25 |
| 10 | Qwen 3 32B | 33.42 |
| 11 | Gemini 2.5 Flash | 37.17 |
| 12 | DeepSeek R1 Distill Qwen 32B | 48.46 |
| 13 | GPT-OSS-20B (Low) | 56.18 |
| 14 | Gemma 3 12B (IT) | 77.33 |
Interactive version: theaggregate.ai/benchmark?slug=phantombench-entities-date · How It Works · Data refreshed daily, snapshot 2026-09-29.