Launching v4.1.1 of the Artificial Analysis Intelligence Index

We have updated the Artificial Analysis Intelligence Index to v4.1.1 - this patch release upgrades our grader models, and brings the latest 𝜏³-Banking version to Artificial Analysis

To keep the Artificial Analysis Intelligence Index the most useful synthesis metric for developers, we make regular updates to the included evaluations and our independent methodology. Today’s update is a minor one to keep our existing evaluation set as reliable as possible.

Overall model rankings remain largely consistent, with a slight increase in scores due to improved grading robustness across the updated evaluations. Claude Opus 5 remains in the #1 position with an Index of 63.

Key changes:

➤ 𝜏³-Banking now runs v1.0.1 from Sierra, updating to the latest upstream task versions and improved grader pipeline that resolves correctness errors in trajectories that recover from unhappy paths

➤ HLE, AA-LCR and AA-Omniscience are now graded by GPT-5.6 Luna (medium), replacing GPT-4o, Qwen3 235B A22B 2507, and Gemini 3 Flash Preview respectively. These checks are now unified under a more capable modern model, selected for strong agreement with human judgment in our grader validation

➤ The effect on scores is small: most models move by less than a point on the Intelligence Index. The largest increase occurred for Muse Spark 1.2 (xhigh, +2.7 points), and the same models hold the top of the leaderboard

Published scores are now using v4.1.1, so all model results on Artificial Analysis now reflect these changes and use our latest consistent, independent methodology.