Today for AI
HOT RADAR
agentHEAT 25.9°

AI CURATED HOT STORY · 11/5/2026

Agent Arena Data Reveals 66% Cost Gap in Long-Horizon Tasks

2 reports archived1 independent sourcesupdated 11/5/2026, 01:49:24
Synthesis & Latest Updates
1 Sources Cross-Validated

Arena.ai released new benchmark data from Agent Arena showing significant cost disparities among frontier models in complex tasks lasting over an hour. Gemini 4 Carbon emerged as the most cost-efficient option with a median hourly cost of $10.51, undercutting Claude Opus 5.5 by 50% and GPT-6 Astra by 66%. While Claude Opus 5.5 leads in capability scores, its high time and monetary costs force developers to re-evaluate the performance-budget trade-off for long-horizon agent workflows.

LATEST/Arena.ai released new benchmark data from Agent Arena showing significant cost disparities among frontier models in complex tasks lasting over an hour. Gemini 4 Carbon emerged as the most cost-efficient option with a median hourly cost of $10.51, undercutting Claude Opus 5.5 by 50% and GPT-6 Astra by 66%. While Claude Opus 5.5 leads in capability scores, its high time and monetary costs force developers to re-evaluate the performance-budget trade-off for long-horizon agent workflows.

Hourly heat curve

2 fully observed hours
Fewer than 3 fully observed hours; trend not plotted yet.

Coverage timeline

Total 2 reports · Latest first
  2. Arena.ai (@arena)T1·68 pts

    Arena Data: Gemini 4 Carbon Scales Cost Better on Complex Tasks

    • Gemini 4 Carbon demonstrates superior cost scalability on high-complexity tasks compared to GPT-6 and Claude Opus series.
    • In the top 5% most expensive sessions on Arena, OpenAI and Anthropic lose frontier share to Google and Xiaomi models.
    • Real-world cases show that selecting cost-efficient models yields significant savings in long-horizon tasks like building Android security apps.
Agent Arena Data Reveals 66% Cost Gap in Long-Horizon Tasks | Today for AI