Phare - Bias Resistance: leaderboard
Giskard’s Phare multilingual safety probe (English, French, Spanish), Bias Resistance module: whether it reproduces stereotypes when the prompt invites one. The metric is a resistance score, so a higher number means the model held the line more often.
Metric: Score (%). Source: phare.giskard.ai. 67 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GPT-4.1 Mini | 88.12 | #355 |
| 2 | Grok 4 Fast (Non-reasoning) | 80.26 | #254 (Grok 4 Fast) |
| 3 | Mistral Small 3.2 | 73.9 | #605 |
| 4 | Llama 4 Maverick | 73.65 | #464 |
| 5 | Claude Haiku 4.5 | 70.66 | #280 |
| 6 | Llama 4 Scout | 67.1 | #656 |
| 7 | DeepSeek V3.1 | 65.17 | #264 |
| 8 | Claude Opus 4.5 | 63.2 | #83 |
| 9 | Mistral Large 3 | 62.72 | #401 |
| 10 | Mistral Medium 3.5 | 60.98 | #353 |
| 11 | DeepSeek V3 | 59.75 | #321 |
| 12 | Qwen 3 8B | 58.64 | #674 |
| 13 | Qwen Plus | 55.7 | #385 |
| 14 | DeepSeek V3 (0324) | 55.26 | #340 |
| 15 | Qwen 3.7 Plus | 54.96 | #92 |
Interactive version: theaggregate.ai/benchmark?slug=phare-bias-resistance · How It Works · Data refreshed daily, snapshot 2026-10-20.