SAGE - Automata
Math · 2025-10-08
The Automata split of SAGE. This child benchmark separates a source-reported subtask or subtrack from the parent aggregate so scores at different grains do not share one benchmark_slug.
Top models (higher is better)
| Model | Score |
|---|---|
| Qwen3-VL-Plus-2025-09-23 | 45.6 |
| Gemini 2.5 Flash | 44.0 |
| Kimi K3 | 44.0 |
| Gemini 2.5 Pro | 43.7 |
| Opus 4.7 | 43.6 |
| Gemini 3 Flash Preview | 43.3 |
| Kimi K2.6 | 42.9 |
| Gemini 3.1 Flash-Lite Preview | 42.1 |
| Opus 4.8 | 41.7 |
| Gemma 4 31B IT | 41.7 |
| GPT-5.4 Mini | 41.3 |
| GPT-5.5 | 40.5 |
| Grok 4.20 | 40.5 |
| Claude Fable 5 | 39.7 |
| GPT-5.6 Sol | 39.7 |
| GPT-5.6 Terra | 38.9 |
| Gemini 3.5 Flash | 38.5 |
| Opus 4.5 | 38.1 |
| Opus 4.6 | 38.1 |
| MiniMax M3 | 37.3 |
| O4 Mini | 37.3 |
| o3 | 36.9 |
| Qwen3.6 Plus (2026-04-02) | 36.5 |
| Qwen3.6 27B | 35.7 |
| Sonnet 4.6 | 35.3 |
| Gemini 2.5 Flash Preview (09-2025) | 35.3 |
| Grok 4.3 | 35.3 |
| Qwen3.5-Flash | 34.9 |
| Sonnet 5 | 34.9 |
| Gemini 3.6 Flash | 34.9 |
| Kimi K2.5 | 34.1 |
| Sonnet 4 | 33.3 |
| Gemini 2.5 Flash-Lite | 33.3 |
| Claude Opus 5 | 32.5 |
| Grok 4.5 | 32.5 |
| Qwen3.7-Plus | 32.5 |
| Gemini 3.1 Pro Preview | 32.1 |
| GPT-5.2 | 32.1 |
| GPT-5.6 Luna | 31.4 |
| Gemini 2.5 Flash-Lite Preview (09-2025) | 31.0 |