All results and scores are simulated.
Simulated
Model leaderboard
Question set v1.0.0 · 1e2a63eb258f · jury of 5 · Jev off. Lower bias magnitude is more centered. These numbers are seeded simulations.
| Model | Grade | Bias | Answers | Hedges | Left / Right | Pro / anti government | Western / non-Western | Religious / secular |
|---|---|---|---|---|---|---|---|---|
| Llama 4 Maverickmeta-llama/llama-4-maverick | A | 7.8 | 98% | 14% | +0.02 | -0.01 | +0.04 | -0.01 |
| gpt-oss-120bopenai/gpt-oss-120b | A- | 8.5 | 96% | 10% | +0.14 | -0.07 | +0.06 | +0.06 |
| Mistral Medium 3.5mistralai/mistral-medium-3-5 | A- | 8.7 | 96% | 16% | -0.03 | 0.00 | +0.06 | -0.09 |
| Gemini 3.8 Flashgoogle/gemini-3.8-flash | A- | 8.8 | 96% | 18% | -0.09 | +0.10 | +0.06 | -0.04 |
| Mistral Large 4mistralai/mistral-large-4-0 | A- | 9.2 | 92% | 19% | -0.06 | -0.04 | +0.11 | -0.16 |
| GPT-6.1 Solopenai/gpt-6.1-sol | A- | 9.5 | 88% | 29% | -0.08 | +0.09 | +0.10 | -0.05 |
| GPT-6 Lunaopenai/gpt-6-luna | A- | 9.6 | 95% | 12% | -0.03 | +0.05 | +0.08 | -0.03 |
| Kimi K3moonshotai/kimi-k3 | A- | 10.1 | 94% | 19% | -0.02 | +0.17 | -0.11 | -0.02 |
| Claude Haiku 5.5anthropic/claude-haiku-5.5 | A- | 10.2 | 94% | 22% | -0.15 | +0.06 | +0.11 | -0.06 |
| DeepSeek V4.1 Flashdeepseek/deepseek-v4.1-flash | A- | 10.5 | 93% | 12% | +0.05 | +0.19 | -0.18 | -0.04 |
| Gemini 3.1 Progoogle/gemini-3.1-pro-preview | A- | 10.6 | 88% | 26% | -0.16 | +0.15 | +0.08 | -0.07 |
| GPT-6 Astraopenai/gpt-6-astra | A- | 11.3 | 87% | 38% | -0.13 | +0.07 | +0.16 | -0.10 |
| Qwen 3.7 Plusqwen/qwen3.7-plus | A- | 11.7 | 88% | 16% | +0.04 | +0.25 | -0.21 | -0.03 |
| Claude Sonnet 5.5anthropic/claude-sonnet-5.5 | A- | 12.9 | 88% | 35% | -0.23 | +0.09 | +0.18 | -0.13 |
| Grok 4.3x-ai/grok-4.3 | A- | 13.0 | 95% | 13% | +0.33 | -0.09 | +0.10 | +0.03 |
| Grok 4.7x-ai/grok-4.7 | A- | 13.5 | 98% | 11% | +0.39 | -0.15 | +0.11 | +0.04 |
| Claude Opus 5.5anthropic/claude-opus-5.5 | B+ | 14.1 | 92% | 32% | -0.27 | +0.11 | +0.20 | -0.11 |
| GLM-5.3z-ai/glm-5.3 | B+ | 14.3 | 82% | 20% | +0.10 | +0.34 | -0.26 | -0.07 |
| DeepSeek V4 Prodeepseek/deepseek-v4-pro | B+ | 14.7 | 84% | 21% | +0.08 | +0.28 | -0.24 | -0.07 |
| Qwen 3.8 Maxqwen/qwen3.8-max-0902 | B+ | 15.3 | 81% | 20% | +0.05 | +0.33 | -0.28 | -0.05 |