PhantomBench - Terms (Place): leaderboard
Metric: Hallucination rate (%): share of answers that do not abstain on the Phantom-T subset (1,060 generated non-existent terms modeled on medical abbreviations, science glossaries and legal terms), 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 | Qwen 3 8B | 3.21 |
| 2 | Qwen 3 14B | 3.3 |
| 3 | Llama 3.1 8B Instruct | 3.58 |
| 4 | Qwen 3 4B | 6.23 |
| 5 | Qwen 2.5 7B Instruct | 6.51 |
| 6 | Qwen 3 1.7B | 9.15 |
| 7 | Gemma 2 9B (IT) | 12.83 |
| 8 | Qwen 3 32B | 14.06 |
| 9 | Gemini 2.5 Pro | 15.75 |
| 10 | Gemini 2.5 Flash | 19.34 |
| 11 | Mistral 7B Instruct (v0.3) | 43.68 |
| 12 | DeepSeek R1 Distill Qwen 32B | 58.97 |
| 13 | GPT-OSS-20B (Low) | 72.68 |
| 14 | Gemma 3 12B (IT) | 85.28 |
Interactive version: theaggregate.ai/benchmark?slug=phantombench-terms-place · How It Works · Data refreshed daily, snapshot 2026-09-29.