OpenAI Web Search ranks 5th on Artificial Analysis Search Index

OpenAI: OpenAI is an AI research and deployment company focused on developing advanced language models and integrated tools for developers. Its Web Search feature provides a native search capability directly within model interactions rather than relying on external API calls. In the news, OpenAI's built-in web_search tool is evaluated on the Artificial Analysis Search Index using GPT-5.6 Luna. GPT-5.6 Luna: GPT-5.6 Luna is an OpenAI language model variant configured for medium reasoning effort in agentic tasks. It serves as the candidate answer model in the benchmark, paired with different search providers to measure uplift. In the news, it powers all setups including the OpenAI Web Search evaluation. Responses API: Responses API is an OpenAI interface that enables models to handle tool calls and responses in a single interaction. It supports built-in tools like web_search without external agent loops. In the news, OpenAI Web Search executes entirely within one Responses API call using the web_search tool. Artificial Analysis: Artificial Analysis runs independent benchmarks and leaderboards for AI models, search providers, and related tools. It maintains the Search Index that evaluates search-enhanced performance across multiple quality metrics. In the news, the organization publishes results placing OpenAI Web Search fifth on its leaderboard. Efficiency: The approach reduces token usage compared with many Search API providers while maintaining competitive benchmark performance. Integration: OpenAI Web Search executes the full search loop internally via its native tool rather than using external agent frameworks. Benchmarking: Artificial Analysis evaluates search providers in an agentic setting using a fixed model across quality, cost, and latency dimensions.