# ARC-AGI-2 — AI model rankings

Puzzle-style tests of abstract reasoning and pattern-finding — the kind of thing people find easy but AIs often struggle with. Higher is better.

11 tracked models have a published ARC-AGI-2 score. Higher is better. Scores are as published at each model's release.

## Ranking

| Rank | Model | Developer | Score | Released |
| --- | --- | --- | --- | --- |
| 1 | GPT-5.5 | OpenAI | 84.6% | Apr 23 2026 |
| 2 | GPT-5.4-Pro | OpenAI | 83.3% | Mar 5 2026 |
| 3 | Gemini 3.1 Pro | Google | 77.1% | Feb 19 2026 |
| 4 | Claude Opus 4.7 | Anthropic | 75.8% | Apr 16 2026 |
| 5 | Gemini 3.5 Flash | Google | 72.1% | May 19 2026 |
| 6 | Claude Sonnet 4.6 | Anthropic | 58.3% | Feb 17 2026 |
| 7 | Grok 4.20 Beta | SpaceXAI | 53.3% | Feb 17 2026 |
| 8 | GPT-5.2 | OpenAI | 52.9% | Dec 11 2025 |
| 9 | Muse Spark | Meta | 42.5% | Apr 8 2026 |
| 10 | Gemini 3.0 Flash | Google | 33.6% | Dec 17 2025 |
| 11 | Gemini 3.0 Pro | Google | 31.1% | Nov 18 2025 |


---

Canonical page: https://aireleasetracker.com/benchmark/arc-agi-2
Full dataset: https://aireleasetracker.com/llms-full.txt · JSON: https://aireleasetracker.com/models.json
Source: AI Release Tracker (https://aireleasetracker.com). Benchmark scores are the figures published by the releasing lab at launch.