ARC Prize

// Announcing: ARC Prize Foundation + General Intuition We’re grateful to share that General Intuition has donated $1,001,337 to support the development of ARC-AGI-4. This support helps ARC Prize Foundation continue building open, rigorous benchmarks that measure the frontier of artificial general intelligence. As always, donations do not come with preferred treatment. We have not tested General Intuition’s models, but welcome the opportunity to do so in the future. Read our testing policy:...

// Analyzing GPT-5.5 & Opus 4.7 with ARC-AGI-3 This week we went through 160 replays and reasoning traces from OpenAI’s GPT-5.5 and Anthropic’s Opus 4.7 attempting novel, long-horizon environments. The scores were just one data point, but the interesting story is how they achieved their score. Today we’re open-sourcing our analysis package: GPT-5.5 - 0.43% (Public Replays) Opus 4.7 - 0.18% (Public Replays) As we compare the runs of GPT-5.5 and Opus 4.7 we’re able to see they failed in...

// ARC-AGI-3 Launches March 25 After creating the world's highest-velocity game studio and 10 months of building, we're shipping ARC-AGI-3 on March 25.We're hosting a live launch party March 25, 2026 in San Francisco. Come join us.There will be V3 games to play, a fireside chat with Francois Chollet and Sam Altman, and 150+ of AI's top builders in the room. The evening will be recorded and shared afterward. Join us live

// ARC-AGI-3 Toolkit Today we’re releasing the ARC-AGI-3 Toolkit. Your agents can now interact with ARC-AGI-3 environments at >2,000 FPS locally. We're open-sourcing the environment engine, 3 human-verified games (AI scores <5%), and human baseline scores. ARC-AGI-3 moves from static puzzles to interactive environments. Agents now take sequential actions inside environments, letting us measure how efficiently they learn, not just whether they succeed. The toolkit open sources the ARC-AGI-3...

// ARC Prize 2025 - Recap and Winners ARC Prize 2025 is official wrapped. While grand prize is still open, we awarded over $125K in prizes to open-source solutions pushing on ARC-AGI-2. Thank you to the 1,455 teams that submitted 15,154 entries and 90 papers. Congratulations to our 1st place teams: Paper Prize: Tiny Recursive Reasoning (Jolicoeur-Martineau) Top Score Prize: NVARC In our full analysis: 6 new winner interviews - conversations with top score and paper winners about how they...

// A North Star To AGI ARC-AGI-3 has been live for 30 days: New agents, new games, and efficient intelligence. ARC-AGI-3 Preview continues to evolve. Over the past month we closed out our first agent preview competition, validated action efficiency as a means to measure adaptation, and opened up three new games to the community. We also hosted the ARC-AGI-3 Preview event, featuring a presentation from ARC Prize President Greg Kamradt and a conversation between François Chollet and Dwarkesh...

// A North Star To AGI Today we're announcing a public preview of ARC-AGI-3 Intelligence is interactive. Traditional benchmarks focus on single-turn tasks: answering a question, labeling an image, or filling in the blank. If we are going to measure human-like general intelligence, we need to go further. Real-world intelligence unfolds over time. It depends on context, memory, and continuous learning. Humans thrive in these messy, multi-step environments. To actually evaluate AI’s general...

// A North Star To AGI A sneak peek at Interactive Reasoning Benchmarks. AGI will be measured through interactive environments. Traditional benchmarks focus on single-turn tasks: answering a question, labeling an image, or filling in the blank. But if we are going to measure human-like general intelligence, we need to go further. Real-world intelligence unfolds over time, with context, memory, and continuous learning. Humans thrive in these messy, multi-step environments. To actually evaluate...

// A North Star To AGI The ARC-AGI-2 Technical Report is now published. ARC-AGI-2 incorporates a newly curated and expanded sets of tasks specifically designed to provide a granular signal for assessing the abstract reasoning and problem-solving capabilities of today's AI systems. These revamped sets of tasks target higher levels of fluid intelligence, demanding more sophisticated generalization, while continuing to target the intersection of what is feasible for humans but still out of reach...

// A North Star To AGI Guillermo Barbadillo has scored 10.1% on ARC Prize 2025. Guillermo won 2nd place in ARC Prize 2024. This year’s competition features ARC-AGI-2 and limits compute to $50 per submission. To read more about Guillermo's 2024 submission, see his amazing write up. ARC-AGI-2 To find out more about ARC-AGI-2, watch Francois describe how he made the dataset, performance testing, and change log. Join ARC Prize 2025