AIGrok 4.7 debuts at No. 16 on Agent Arena’s leaderboard
Arena reports a +3.96% net improvement score for Grok 4.7 (xHigh), with a median cost of $1.14 per task versus $0.74 for Grok 4.6 (High).
TLDR
Arena’s September 25, 2026 results put Grok 4.7 (xHigh) at No. 16 overall with a +3.96% net improvement score. It ranked No. 6 for Confirmed Success but No. 28 for Steerability. Arena reported a $1.14 median cost per task, compared with $0.74 for Grok 4.6 (High) and $0.43 for Grok 4.5. Arena says its leaderboard measures outcomes relative to the average model across millions of real-world, long-horizon agent tasks, where models can use web search, filesystem and terminal tools.
Combined views
42.8K
5 Sources, first seen 16h ago
Grok 4.7 debuts at No. 16 on Agent Arena’s leaderboard
Arena reports a +3.96% net improvement score for Grok 4.7 (xHigh), with a median cost of $1.14 per task versus $0.74 for Grok 4.6 (High).
TLDR
Arena’s September 25, 2026 results put Grok 4.7 (xHigh) at No. 16 overall with a +3.96% net improvement score. It ranked No. 6 for Confirmed Success but No. 28 for Steerability. Arena reported a $1.14 median cost per task, compared with $0.74 for Grok 4.6 (High) and $0.43 for Grok 4.5. Arena says its leaderboard measures outcomes relative to the average model across millions of real-world, long-horizon agent tasks, where models can use web search, filesystem and terminal tools.