Report
Gemini 4 Carbon (High) lands No. 1 in Text Arena, No. 2 in Agent Arena and No. 4 in Code Arena: WebDev
Arena reports a $1.80 median cost per Agent Arena task, 72% less than top-ranked Claude Opus 5.5 (Max).
TLDR
Arena says Google DeepMind’s Gemini 4 Carbon (High) debuted at No. 1 in Text Arena with 1,528 points, No. 2 in Agent Arena and No. 4 in Code Arena: WebDev with 1,768 points. Arena reports a 16.2% net improvement across 21.9K-plus real-world agentic sessions in Agent Arena.
Combined views
2 Sources, first seen ago
reposts
