https://romainfrog.github.io/ Research notes on AI agents, evaluation environments, and agent benchmarks by Romain Froger. en Mon, 07 Sep 2026 00:00:00 +0200 https://romainfrog.github.io/blog/omnilingual-gaia2/ https://romainfrog.github.io/blog/omnilingual-gaia2/ Wed, 12 Aug 2026 00:00:00 +0200 A machine-translated expansion of four GAIA2 capabilities, evaluating seven AI agents across ten languages beyond English. https://romainfrog.github.io/blog/gaia2-cli/ https://romainfrog.github.io/blog/gaia2-cli/ Wed, 15 Apr 2026 00:00:00 +0200 Gaia2-CLI adds first-class terminal support to the Gaia2 benchmark for evaluating agentic systems in real execution environments.