TranscendPlexity: 540/540 ARC-AGI-1/2/3, 13 tasks with 0% AI solve rate, solved
TranscendPlexity has successfully solved 13 ARC-AGI-2 tasks that were previously deemed impossible for any AI system. These tasks had a 0% solve rate across all known AI attempts, including major models like GPT-4 and Claude. The achievement highlights a significant advancement in AI capabilities through the use of deterministic Python code without dependencies on machine learning models.
- ▪TranscendPlexity solved all 13 ARC-AGI-2 tasks, achieving a perfect score.
- ▪No other AI systems, including GPT-4 and Gemini, were able to solve these tasks.
- ▪The solvers were created using LLM-guided program synthesis, resulting in readable and deterministic code.
Opening excerpt (first ~120 words) tap to expand
🔓 13 "Impossible" ARC-AGI-2 Tasks — All Solved These 13 ARC-AGI-2 evaluation tasks have never been solved by any AI system — not GPT-4, not Claude, not Gemini, not NVARC, not MindsAI, not any Kaggle submission. They have a 0% AI solve rate across all publicly tracked attempts. TranscendPlexity solved all 13. The Scoreboard System Tasks Solved (of 13) Overall ARC-AGI-2 Score TranscendPlexity 13 / 13 ✅ 120 / 120 (100%) NVARC (Kaggle 1st) 0 / 13 24% The ARChitects (2nd) 0 / 13 16.5% MindsAI (3rd) 0 / 13 12.6% GPT-4o 0 / 13 9% Claude 3.5 Sonnet 0 / 13 21% Gemini 1.5 0 / 13 8% Source: ARC Explainer — Unsolved Puzzles The 13 Tasks Task ID Solver Lines Train/Test Pairs Status abc82100 239 4 / 1 ✅ Solved 21897d95 525 4 / 2 ✅ Solved e12f9a14 348 4 / 2 ✅ Solved a32d8b75 303 3 / 2 ✅ Solved 9bbf930d…
Excerpt limited to ~120 words for fair-use compliance. The full article is at GitHub.