Relace: Relace Search

The Relace Search model is engineered for highly efficient and precise codebase exploration. Unlike traditional RAG, it leverages agentic multi-step reasoning, utilizing 4-12 `view_file` and `grep` tools in parallel to identify and return relevant files based on user requests. This advanced approach allows it to achieve results 4x faster than leading frontier models. Relace Search is specifically designed to function as a subagent, passing its findings to an 'oracle' coding agent that then orchestrates and performs the remainder of the coding task. To integrate this model, users need to build an appropriate agent harness and parse the response for information to hand off to the oracle. Detailed documentation is available on the Relace website. With a generous 256K token context window and a maximum output of 4K tokens, Relace Search offers robust capabilities for complex tasks. It supports functions, streaming, and search, making it best suited for chat-based interactions. Pricing is competitive at $1.00 per 1M input tokens and $3.00 per 1M output tokens, available on the PRO access tier. Note that this model does not support image generation.