Arena

Arena, formerly LMArena / Chatbot Arena, is a model-output comparison and evaluation platform with preference voting, agent-category performance and cost views, and a Code Arena development mode.

Scope described by official sources

• Agent documentation describes Code/Chat/Work performance and task-cost views, plus output-token and efficiency views; a limited task window reduces context carry-over effects
• The 2026-07-02 Code Arena article describes databases, authentication, external integrations and deployment; development modes and leaderboard observations have different scopes

Suggested starting workflow

• Define your task category, then read the methodology, sample and cost definition
• Record overall ranking, category, cost percentiles and candidate models separately instead of using one ranking for every task
• Prepare lawful non-sensitive inputs and review platform/model terms before actual use; record conditions, results and cost separately

These are editorial starting suggestions, not a workflow tested by this site.

Before you use it

• The official About page establishes name continuity. Arena-Rank examples are not current live rankings and do not replace the hosted interface or evaluations executed by this site
• Platform task costs reflect particular sessions, segmentation and model conditions, not full same-plan task costs measured here; token rates alone cannot price an entire task
• Terms allow input transfer to third-party model providers and include content-license/data-processing scope. This is not assumed to be a confidential workspace, and accuracy or uniqueness is not guaranteed
• Terms restrict automated queries/collection and manipulated votes; no further platform requests followed this reading. No login, voting, leaderboard-data downloads, app generation or model task took place. Full price, eligibility, commercial rights and model conditions remain unverified