https://cornball.ai/tags/evals/Recent content in Evals on cornball.aiHugo -- gohugo.ioenTue, 15 Sep 2026 11:09:13 -0500https://cornball.ai/posts/corteza-arc-agi-3/Mon, 14 Sep 2026 00:00:00 +0000https://cornball.ai/posts/corteza-arc-agi-3/I’m pleased to announce that corteza has scored 98.58% on the ARC-AGI-3 challenge with Claude Opus 5. corteza is an R-native agent harness built around a persistent R session: functions, data, and intermediate work survive across model calls, and even across context compaction. wa30, a Sokoban-style carry game and one of the 23 games corteza finished at a score of 100. The rules are unknown by design; the model has to work them out.