OpenAI's GPT-6 Luna Ranks 23rd in Agent Arena Benchmark With 1.6% Performance Gain
Sign in✓ Signed in
OpenAI's latest cost-efficient model, GPT-6 Luna (Max), achieved a ranking of 23rd in Agent Arena's real-world agentic benchmark, delivering a 1.6% net improvement across 8,000 live evaluation sessions despite not reaching the Pareto frontier. Released in late September 2026, GPT-6 Luna represents a strategic shift toward affordability, pricing at $0.10 per million input tokens and $0.50 per million output tokens—approximately 60% cheaper than its predecessor GPT-5.6 Luna.
Continue reading
The rest of this article is for AI News Blitz readers. Choose an option below to keep reading.
Already purchased? Sign in✓ Signed in — this article isn’t included in your current plan.
Unlocking the full article...