Antidote: Everyday Edition
Safety · 2026-05-21
Elo rating on real-world everyday prompts graded by experts.
Top models (higher is better)
| Model | Score |
|---|---|
| Claude Fable 5 | 1099 |
| Gemini 3.1 Pro Preview | 1091 |
| Gemini 3.8 Flash | 1087 |
| Gemini 3.6 Flash | 1083 |
| Gemini 3.5 Flash | 1079 |
| Kimi K3 | 1077 |
| Gemini 3.7 Flash | 1075 |
| GLM-5.3 | 1067 |
| Claude Fable 5.1 | 1065 |
| Claude Opus 5 | 1055 |
| Qwen3.7-Max (source unspecified) | 1051 |
| Kimi K2.6 | 1045 |
| Opus 4.6 | 1042 |
| Opus 4.7 | 1042 |
| GLM-5.2 | 1040 |
| Opus 4.8 | 1030 |
| Muse Spark 1.2 | 1028 |
| GPT-5.6 Sol | 1027 |
| Hy3 | 1022 |
| DeepSeek V4 Flash Vision Exp | 1015 |
| Muse Spark 1.1 | 1014 |
| GPT-5.5 | 1013 |
| Sonnet 4.6 | 1012 |
| DeepSeek-V4-Pro | 1007 |
| Kimi K2.5 | 1006 |
| Grok 4.20 | 1001 |
| Qwen3.5 Plus (2026-02-15) | 999 |
| Gemini 3.5 Flash-Lite | 991 |
| Inkling | 991 |
| Grok 4.5 | 973 |
| DeepSeek-V4-Flash | 966 |
| Grok 4.6 | 966 |
| DeepSeek-V3.2 | 954 |
| Grok 4.3 | 950 |
| ERNIE 5.1 | 947 |
| Mistral Large 3 675B Instruct 2512 | 947 |
| Haiku 4.5 | 932 |
| Muse Glimmer 30B | 931 |
| GPT-5.4 Mini | 929 |
| Gemma 3 12B IT | 893 |