Arena has turned on a live Pareto frontier view in Agent Arena, plotting net improvement on real-world agentic tasks against median cost per task so buyers can see which models sit on the cost-performance edge rather than on a single ranking. As of the Aug. 18, 2026 snapshot, the board covers 1,878,054 sessions and 49 models, and the models Arena lists as Pareto-optimal are Claude Opus 5 (High) at +12.34% and $1.78 per task, Kimi K3 (Max) at +10.53% and $0.62, GPT 5.5 (High) at +7.75% and $0.44, GPT 5.5 at +6.40% and $0.28, Grok 4.5 at +6.08% and $0.22, GLM 5.2 (Max) at +6.06% and $0.18, GPT 5.6 Luna (xHigh) at +4.25% and $0.04, and Mimo V2.5 Pro at -2.22% and $0.03.
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...
Contact us
Questions, corrections, or ideas? Send us a message here. We aim to reply by email within 24 hours.
Thank you. Your message has been sent. We aim to reply by email within 24 hours.Please enter your email address and a message.We could not send your message. Please try again in a little while.