models · release 2026-09-18
Models
Every model on the boards, scored on the same questions against human labels. Decision Score: 100 = perfect, 0 = guessing the label base rates.
| Model | Type | noul | choice | score | |
|---|---|---|---|---|---|
| Gemini 3.8 FlashGoogle | LLM | 73.0tied for rank 1 of 8 | 74.1rank 1 of 8 | 4.6tied for rank 1 of 8 | vs Jev |
| JevTypeSafe AI | System One | 69.0tied for rank 1 of 8 | 67.8tied for rank 2 of 8 | 9.2tied for rank 1 of 8 | |
| GLM-5.3Z.ai | LLM | 60.6tied for rank 3 of 8 | 66.8tied for rank 2 of 8 | 7.8tied for rank 1 of 8 | vs Jev |
| DeepSeek V4.1 FlashDeepSeek | LLM | 47.5rank 7 of 8 | 63.9tied for rank 4 of 8 | −19.0rank 7 of 8 | vs Jev |
| Qwen3.8 FlashAlibaba | LLM | 62.4rank 2 of 8 | 62.4tied for rank 4 of 8 | −1.4tied for rank 3 of 8 | vs Jev |
| Mistral Medium 3.5Mistral AI | LLM | 58.0tied for rank 3 of 8 | 59.5rank 5 of 8 | −13.7rank 5 of 8 | vs Jev |
| Mercury 2.5Inception | LLM | 55.7tied for rank 3 of 8 | 54.0rank 7 of 8 | −5.5tied for rank 3 of 8 | vs Jev |
| Label prior | Baseline | 0.0rank 8 of 8 | 0.0rank 8 of 8 | 0.0tied for rank 1 of 8 |