# Two API settings triple GPT-5.6 ARC-AGI-3 scores

> Enabling reasoning retention and compaction settings in the GPT-5.6 API dramatically improved benchmark performance on ARC-AGI-3, showing that configuration choices can significantly affect model capability on abstract reasoning tasks.

| | |
|---|---|
| **Tool** | OpenAI |
| **Version** | — |
| **Kind** | announcement |
| **Published** | 2026-07-29 |
| **Observed** | 2026-08-11 |
| **Significance** | 3/5 |
| **Breaking** | no |
| **Categories** | capability |

## What changed


Two API settings for GPT-5.6 — retaining reasoning traces and enabling compaction — together tripled scores on the ARC-AGI-3 benchmark while also improving efficiency.


## Sources

- [blog_rss](https://openai.com/index/how-two-settings-tripled-our-arc-agi-3-scores) — retrieved 2026-08-11


## Community

_No curated reactions recorded. Facts and community takes are kept in separate layers
and never blended._

---
Canonical: https://changelogs.info/openai/two-api-settings-triple-gpt-5-6-arc-agi-3-scores
Entity: https://changelogs.info/openai
Event ID: `evt_2026-07-29_openai_how-enabling-two-settings-tripled-our-scores-on-the-arc-agi-3-benchmark`
Licence: event synthesis © changelogs.info, CC BY 4.0. Linked sources belong to their vendors.