AI evaluation platform with OpenAI Responses API and built-in web search tool support. Test prompts against datasets, score with LLM judges, and trace MCP tool calls. Works with Claude, OpenAI, and Azure.