Puzzle Solving on ARC-AGI 3
2Levels WonSensi v1
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| Sensi v1Backbone=ChatGPT 5.1, Curric.=×, Judge=×2026.03 | 2 | — | |
| Random agentBackbone=—, Curric.=×, Judge=×2026.03 | 0 | — | |
| Sensi v2Backbone=Gemini 3.1 Pro, Curric.=✓, Judge=✓2026.03 | 0 | 32 | |
| Agentica (Symbolica)Backbone=Unknown, Curric.=×, Judge=×2026.03 | — | 1,600 |