ARC-AGI-2
The short answer · 10 words
In short
Successor of ARC-AGI-1 with more demanding tasks for rule-based generalization.
Flow-alerta.cab (2026): ARC-AGI-2. https://flow-alerta.cab/en/glossar/arc-agi-2. Editorially reviewed on: 09 October 2026.
How much detail would you like?
Practical
Helps to classify progress and limitations in novel problem-solving tasks.
Example scenario: A model is tested with previously unknown grid tasks under a fixed computing budget.
Relevant for research and benchmark understanding; does not replace operational evaluation.
At a glance
- Core
- Like ARC-AGI-1, ARC-AGI-2 uses input and output grids in which a new transformation is to be derived from examples. The tasks have been made more demanding. Results are tied to tasks, test conditions and computing budget.
- Use
- Helps to classify progress and limitations in novel problem-solving tasks.
- Example
- Example scenario: A model is tested with previously unknown grid tasks under a fixed computing budget.
- Limits
- Compare scores of different test setups directly; equate success with general intelligence; use test data for training.
In context
Questions about this term
Is this the same set of tasks as ARC-AGI-1?
No.
Does a solved task prove AGI?
No.
All details and background
Like ARC-AGI-1, ARC-AGI-2 uses input and output grids in which a new transformation is to be derived from examples. The tasks have been made more demanding. Results are tied to tasks, test conditions and computing budget.
ELI5 – Explained simply
The picture puzzles remain similarly structured, but require more difficult combinations of rules.
Why it matters in practice
Helps to classify progress and limitations in novel problem-solving tasks.
Example
Example scenario: A model is tested with previously unknown grid tasks under a fixed computing budget.
Pitfalls and limits
- Compare scores of different test setups directly; equate success with general intelligence; use test data for training.
Origin & history
ARC Prize published ARC-AGI-2 in March 2025; the technical report explains the selection of more demanding tasks.
Use in mid-sized companies
- Relevant for research and benchmark understanding; does not replace operational evaluation.
Related terms
Matching Flow-alerta service
AI adoption & supportAI introduction and accompaniment.Matching tool / use case
View use casesUse Case: Comparison of problem-solving procedures under documented test conditions.Sources
- arcprize.orgARC Prize, Einführung
- arcprize.orgARC Prize, technischer Bericht
Quellenprüfung: 09.10.2026. Definition und dokumentierte Einordnung abgeglichen; Beispiele beschreiben mögliche Anwendungen.
Editorially reviewed on: