GPT-6 Luna (Max) reaches Arena.ai’s Agent Arena Pareto frontier

Arena.ai says GPT-6 Luna (Max) has landed on the Agent Arena Pareto frontier with a +1.59% net improvement and a median cost of $0.05 per task. Earlier the same day, Arena.ai had ranked the model 23rd with a +1.6% net improvement across 8,000 real-world agentic sessions and said it had not reached the frontier.

Arena.ai says Luna costs 29% less than DeepSeek-V4.1-Flash and 76% less than Hy4 preview, while its net-improvement score trails those models by 2.27 and 2.51 percentage points. Among other GPT models on the frontier, it costs 94% less than GPT-6 Sol (Max) and 98% less than GPT-6 Astra (Max).