OpenAI agents now default to the Responses API

The openai: model prefix now resolves to OpenAIResponses rather than the legacy Chat Completions surface. New agents written as Agent(model="openai:...") automatically route through OpenAI's Responses API, giving teams access to its richer feature set, including built-in tools and improved streaming behavior, without code changes.

Action required: If your application depends on Chat Completions semantics, switch to the new openai-chat: prefix, for example Agent(model="openai-chat:gpt-4.1"). No action is needed for teams that already instantiate OpenAIChat or OpenAIResponses directly.

Why it matters: Aligning the default with the Responses API moves new agents onto OpenAI's actively developed surface, reduces friction when adopting newer capabilities, and makes the most capable behavior the path of least resistance for everyone building on Agno.