Did Codex Reset
GitHub

Arena.ai ranks Grok 4.7 (xHigh) 16th in Agent Arena at +3.96%

Arena.ai

Arena.ai ranks SpaceXAI’s Grok 4.7 16th in Agent Arena, with a net improvement score of +3.96%. It says Grok 4.7 (xHigh) improves on earlier versions while costing more per task than Grok 4.6 (High).

Median cost and net improvement are $1.14 per task and +3.96 for Grok 4.7 (xHigh), $0.74 and +1.22 for Grok 4.6 (High), and $0.43 and +1.50 for Grok 4.5.

By signal, Arena.ai ranks Grok 4.7 (xHigh) sixth for Confirmed Success (+10.41%), 12th for Bash Recovery (+5.98%), 16th for Praise vs Complaint (+4.93%), and 28th for Steerability (−1.89%). It also reports no Tool Hallucination issues, at +0.36%.