Phare - Hallucination Resistance: leaderboard
Giskard’s Phare multilingual safety probe (English, French, Spanish), Hallucination Resistance module: whether the model corrects a misleading premise instead of elaborating on it. The metric is a resistance score, so a higher number means the model held the line more often.
Metric: Score (%). Source: phare.giskard.ai. Saturation forecast: Around January 2027. 71 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Claude Opus 4.5 | 88.23 | #83 |
| 2 | Claude Opus 4.6 | 87.76 | #63 |
| 3 | Gemini 3.1 Pro (Preview) | 87.59 | #58 |
| 4 | Claude Sonnet 4.5 | 87 | #142 |
| 5 | Claude 3.5 Sonnet | 86.71 | #350 |
| 6 | Claude Opus 4.1 | 86.19 | #140 |
| 7 | Claude Sonnet 5 | 86.1 | #57 |
| 8 | Claude Sonnet 4.6 | 85.51 | #90 |
| 9 | Claude 3.7 Sonnet | 85.17 | #249 |
| 10 | Claude Haiku 4.5 | 83.56 | #280 |
| 11 | Kimi K2.6 | 82.78 | #104 |
| 12 | Gemini 3.5 Flash | 82.18 | #60 |
| 13 | DeepSeek V4 Flash | 81.88 | #137 |
| 14 | GPT-5.1 | 81.84 | #135 |
| 15 | Qwen 3.7 Max | 81.14 | #75 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=phare-hallucination-resistance · How It Works · Data refreshed daily, snapshot 2026-10-20.