Claude Haiku 5.5 hits #27 on Agent Arena at $0.09 per task, 86% cheaper than GPT-5.4
arena · x · 2026-10-14
Arena released real-world results for Anthropic's Claude Haiku 5.5 (High): across 20.5K+ agentic sessions it scored +1.48% net improvement, ranking #27 overall and #15 in Code (+4.83%).
Cost is the headline:
- Median cost of $0.09/task, just outside the Pareto frontier; GPT-6 Luna (Max) offers +3.19% at $0.08/task
- 86% cheaper than GPT-5.4 (High) at $0.63/task in the #26 spot
- 97–99% cheaper per task than Claude Sonnet 5.5 (Max), Fable 5.1 (Max), and Opus 5.5 (Max)
Earlier Code Arena (WebDev) debut: #30 at 1587 pts, priced $0.10/$0.50 per 1M input/output tokens — matching GPT-6 Luna while scoring six points higher, and 90% cheaper than Haiku 4.5.
More from coding & agent
- Google's ADK brings one agent framework to Python, Go, TypeScript, Java and Kotlin — rseroter · 2026-10-15
- AI agent flags a bizarre '71 >= 71' error while a server pool balloons by 1.4TB — haydendevs · 2026-10-15
- Invideo launches an MCP for editing videos directly via Claude or ChatGPT — azed_ai · 2026-10-15
- Connecting Grok to ROS via OpenClaw and AgenticROS for robotics missions — chrismatthieu · 2026-10-15
- Automating bug report fixes: a Bot triages DMs into a Cursor project — msfeldstein · 2026-10-15
- Devin now auto-fixes Pylon support tickets; new customers get $15,000 in credits — DevinAI · 2026-10-15