An abstract-reasoning benchmark created by the founder of Keras to evaluate progress toward general intelligence.
Browse the latest scores, model modes, release dates, and parameter sizes for ARC-AGI.
Data sourced primarily from official releases (GitHub, Hugging Face, papers), then benchmark leaderboards, then third-party evaluators. Learn about our data methodology
1 parallel-mode results hidden
| Rank | Model | License | |||
|---|---|---|---|---|---|
 GPT-5.5 Pro Thinking Level · High | 96.50 | 2026-04-23 | Unknown | Closed | |
GPT-5.6 Terra Thinking Level · Max | 96.50 | 2026-06-26 | Unknown | Closed | |
 Gemini 3 Deep Think - 2620 Thinking Enabled | 96.00 | 2026-02-13 | Unknown | Closed | |
4 | GPT-5.5 Thinking Level · Extra High | 95.00 | 2026-04-23 | Unknown | Closed |
5 | GPT-5.5 Pro Thinking Level · Extra High | 95.00 | 2026-04-23 | Unknown | Closed |
6 | GPT-5.4 Pro Thinking Level · High | 94.50 | 2026-03-05 | Unknown | Closed |
7 | GPT-5.5 Thinking Level · High | 94.50 | 2026-04-23 | Unknown | Closed |
8 | GPT-5.4 Standard Mode | 93.70 | 2026-03-05 | Unknown | Closed |
9 | GPT-5.4 Thinking Level · Extra High | 93.70 | 2026-03-05 | Unknown | Closed |
10 |  Opus 4.7 Thinking Level · High | 93.50 | 2026-04-16 | Unknown | Closed |
11 | GPT-5.5 Thinking Level · Medium | 92.20 | 2026-04-23 | Unknown | Closed |
12 | Claude Opus 4.6 Extended Thinking | 92.00 | 2026-02-05 | Unknown | Closed |
13 | Opus 4.7 Thinking Level · Max | 92.00 | 2026-04-16 | Unknown | Closed |
14 | Opus 4.7 Thinking Level · Low | 91.00 | 2026-04-16 | Unknown | Closed |
15 | Opus 4.7 Thinking Level · Medium | 91.00 | 2026-04-16 | Unknown | Closed |
16 | GPT-5.2 Deep Thinking Mode | 90.50 | 2025-12-11 | Unknown | Closed |
17 | GPT-5.2 Pro Thinking Enabled | 90.50 | 2025-12-11 | Unknown | Closed |
18 | GPT-5.6 Luna Thinking Level · Max | 88.00 | 2026-06-26 | Unknown | Closed |
19 | GPT-5.2 Thinking Level · Extra High | 86.20 | 2025-12-11 | Unknown | Closed |
20 | GPT-5.4 Thinking Level · Medium | 86.20 | 2026-03-05 | Unknown | Closed |
21 | Claude Opus 4.6 Thinking Level · Low | 86.00 | 2026-02-05 | Unknown | Closed |
22 | Opus 4.5 Extended Thinking | 80.00 | 2025-11-25 | Unknown | Closed |
23 | GPT-5.2 Thinking Level · High | 78.70 | 2025-12-11 | Unknown | Closed |
24 | GPT-5.5 Thinking Level · Low | 76.20 | 2026-04-23 | Unknown | Closed |
25 | Gemini 3.0 Pro (Preview 11-2025) Thinking Enabled | 75.00 | 2025-11-18 | Unknown | Closed |
26 | GPT-5.1 Thinking Level · High | 72.80 | 2025-11-12 | Unknown | Closed |
27 | GPT-5.2 Thinking Level · Medium | 72.70 | 2025-12-11 | Unknown | Closed |
28 | GPT-5-Pro Thinking Enabled | 70.20 | 2025-08-07 | Unknown | Closed |
29 | GPT-5.4 Thinking Level · Low | 68.20 | 2026-03-05 | Unknown | Closed |
