Vibeleaderboard
← All Intel
Intel / video

Build Hour: Responses API

Source
youtube.com
Author
OpenAI
Date

Context

In a Build Hour from OpenAI, an API engineer explains why the company introduced the Responses API alongside Chat Completions. Chat Completions samples the model once per request. Responses is built as an : in one request the model can write code, have it run on the server, see the result and continue until done. Its output is a list of typed items (message, reasoning, function call, call), where Chat Completions treated everything as a message.

A central design point is that reasoning can be passed from one request to the next, so after each tool call the model can carry on with its plan instead of thinking again from scratch. The engineer reports a 5% gain on OpenAI's primary tool-calling evaluation, and that long multi-tool runs are 20% faster at the median and use fewer output . These are OpenAI's figures, and the session does not show how they were measured.

Terms in this piece · Glossary
  • agentic loop — The cycle an agent runs in: decide, call a tool, read the result, decide again — repeating until the goal is met or a stop condition fires.
  • MCP — The Model Context Protocol — an open standard that lets any AI assistant plug into any tool or data source without custom integration code.
  • token — The chunk of text a model reads and writes in — roughly three-quarters of a word — and the unit AI usage is billed in.
  • AI agent — An AI system that doesn't just answer once but works toward a goal in a loop — taking actions, reading the results, and deciding what to do next.
Read the source www.youtube.com
More from OpenAI
Recommended reads
Comments

Checking sign-in…

Loading comments…