New agent framework hits 85.1% on ARC AGI3 public games
JFPuget · x · 2026-07-27
A new agent framework from the poster’s colleagues reached 85.1% on the ARC AGI3 public games, according to the attached chart.
- The chart compares NOOA + GPT-5.6-sol with world model + memory against several baselines.
- NOOA + GPT-5.6-sol: 85.1% fleet-mean RHAE after 120 minutes.
- NOOA + GPT-5.5 world model + memory: 50.2%.
- NOOA + GPT-5.5 baseline skill: 41.7%.
- NOOA + GPT-5.5 markdown files: 38.4%.
The post says the work is by Elad Sarafian and Gal Kaplun, with the author’s team contributing a bit to it.
Related event: New Agent Framework Scores 85.1% on ARC-AGI3(4 posts)→
More from coding & agent
- Google Cloud rounds up agentic AI workflow tutorials and serverless builds — rseroter · 2026-07-27
- New agent framework reaches 85% on ARC AGI3 public games — JFPuget · 2026-07-27
- Millions of Agents: The Challenge of Locating and Remotely Shutting Down Rogue Instances — jonasgeiping · 2026-07-27
- Anthropic engineer shares a 2-hour workshop on graph engineering for agents — usamawahabkhan · 2026-07-27
- LangChain’s Interrupt workshops focus on deep agents, production and monitoring — LangChain · 2026-07-27
- Cohere launches North Automations for plain-language enterprise workflows — cohere · 2026-07-27