OpenCodePapers

abstract-reasoning-on-arc-agi

Visual ReasoningAbstract Reasoning
Dataset Link
Results over time
Click legend items to toggle metrics. Hover points for model names.
Leaderboard
PaperCodeAccuracyModelNameReleaseDate
ARC Prize Leaderboard✓ Link98.5Claude Fable 5 (Max)2026-06-09
ARC Prize Leaderboard✓ Link98.5Gemini 3.8 Flash (High)2026-09-02
GPT-6 Astra: A new generation of intelligence✓ Link98.5GPT-6 Astra2026-09-03
ARC Prize Leaderboard✓ Link98.5Claude Opus 5.5 (High)2026-09-22
ARC Prize Leaderboard98.5GPT-6.1 Sol (XHigh)2026-09-29
ARC Prize Leaderboard✓ Link98.5Gemini 4 Argon (High)2026-09-30
ARC Prize Leaderboard98.0Gemini 3.1 Pro (Preview)2026-02-19
ARC Prize Leaderboard✓ Link97.5GPT-5.6 Sol (XHigh)2026-07-09
Introducing Claude Opus 5✓ Link97.5Claude Opus 52026-07-24
ARC Prize Leaderboard97.5Claude Fable 5.1 (Max)2026-09-01
ARC Prize Leaderboard96.5GPT-5.5 Pro (High)2026-04-23
ARC Prize Leaderboard96.0Gemini 3 Deep Think (2/26)2025-11-18
ARC Prize Leaderboard95.5GPT-6 Sol (Max)2026-09-22
Introducing GPT-5.494.5GPT-5.4 Pro2026-03-05
ARC Prize Leaderboard✓ Link94.5Kimi K3 (Max)2026-07-27
ARC Prize Leaderboard✓ Link94.5DeepSeek V4.1 Flash (Max)2026-09-17
ARC Prize Leaderboard94.0Claude Opus 4.6 (120K, High)2026-02-05
ARC Prize Leaderboard93.5Claude 4.7 (High)2026-04-16
ARC Prize Leaderboard92.5Gemini 3.5 Flash (High)2026-05-19
ARC Prize Leaderboard92.5Claude Opus 4.8 (Max)2026-05-28
ARC Prize Leaderboard91.167Gemini 3.6 Flash (High)2026-07-21
ARC Prize Leaderboard✓ Link91.0GLM-5.3-Flash (Max)2026-08-26
Introducing GPT-5.290.5GPT-5.2 Pro2025-12-11
optimize_anything: A Universal API for Optimizing any Text Parameter✓ Link89.5optimize_anything ARC-AGI agent (Gemini 3 Flash)2026-05-19
ARC Prize Leaderboard✓ Link89.0DeepSeek V4 Flash 0731 (Max)2026-04-26
ARC Prize Leaderboard89.0Claude Haiku 5.5 (Max)2026-10-07
ARC Prize Leaderboard87.5Grok 4.6 (Medium)2026-08-12
Inkling-Small✓ Link84.0Inkling-Small2026-07-27
ARC Prize Leaderboard✓ Link79.5Inkling2026-07-15
Natural Image Pretraining Improves Abstract Reasoning70.2Nat-ARC ensemble2026-09-08
ARC Prize Leaderboard70.17GPT-5 Pro2025-12-19
ARC Prize Leaderboard✓ Link65.33Kimi K2.52026-02-02
ARC Prize Leaderboard63.67Claude Sonnet 4.5 (Thinking 32K)2025-09-29
ARC Is a Vision Problem!60.4VARC (ensemble)2025-11-18
SSL: Sweet Spot Learning for Differentiated Guidance in Agentic Optimization58.2SSL (7B)2026-01-30
ARC Prize Leaderboard✓ Link57DeepSeek V3.22025-12-02
ARC Is a Vision Problem!54.5VARC2025-11-18
Human-like Abstract Visual Reasoning via Understanding and Solving Reasoning Loop47.2USRL (Ours)2026-06-03
Hierarchical Reasoning Model✓ Link40.3HRM2025-06-26
ARC Prize Leaderboard40.0Claude Sonnet 4 (Thinking 16K)2025-05-22
ARC Prize Leaderboard✓ Link40.0Tiny Recursion Model (TRM)2025-10-06
ARC Prize Leaderboard28.6Claude 3.7 (16K)2025-02-24
ARC Prize Leaderboard✓ Link21.21Deepseek R1 (05/28)2025-01-22
ARC Prize Leaderboard✓ Link11.0Qwen3-235b-a22b Instruct (25/07)2025-05-14
ARC Prize Leaderboard6.12Magistral Medium (Thinking)2025-06-12
ARC Prize Leaderboard✓ Link4.38Llama 4 Maverick2025-04-05