Audience
AI application developers, routing developers, and engineering teams building AI applications and agents, including teams using hosted, hybrid, or local inference.
About Rayline ARC
Rayline ARC is an adaptive large language model (LLM) router and inference API for developers building AI applications and agents. It evaluates each request in conversation context and selects a model and reasoning level using a routing policy shaped by the workload’s success value, exploration settings, and model prices. Developers can use hosted ARC models that return generated responses, or call the Routes API to receive a model recommendation and run inference through their own endpoints. ARC offers preset tiers for everyday and demanding tasks, a configurable quality-cost option, and supported connections for coding agents and frameworks including Claude Code, Codex, Cursor, OpenCode, and LangGraph. Teams can also run ARC’s routing model on their own infrastructure, connect local models, and review usage and charges.