Urgent.News

What's breaking now, across thousands of outlets.

AI

OpenAI DevDay 2026 Recap for Developers

OpenAI announced a series of product and developer updates at DevDay 2026, including GPT-6.1 Sol, computer use for the Agents API, cloud-based Codex environments, a Decisions API, and new plugin capabilities for ChatGPT. By Daniel Dominguez

OpenAI's DevDay 2026 conference featured a number of updates aimed at developers. New capabilities were introduced for the Agents API, allowing applications to operate software through graphical interfaces and supporting multi-agent functionality. GPT-6.1 Sol, an updated version of GPT-6 Sol, was released with improved coding, computer use, and professional task capabilities.

This model was noted to perform nearly as well as GPT-6 Astra on certain evaluations, while costing significantly less. Codex now has the ability to run in cloud environments, enabling developers to initiate remote tasks from various devices. The Codex CLI was also updated with voice input and a /agents interface. A Decisions API was introduced, which uses the smaller Luna model to select answers based on text or image context.

ChatGPT's plugin system expanded with new features, including sidebar experiences, interactive conversation panels, and custom file viewers. Dots, persistent agents, and ChatGPT Space, a shared workspace, were also unveiled. These changes aim to shift OpenAI's platform from individual model calls to persistent agent workflows, capable of tool usage, software operation, collaboration, and remote task execution.

Community discussions centered around the growing focus on agents and developer infrastructure, with developers praising features like computer use, cloud-based Codex, and lower-cost models. However, some questioned whether these releases represented significant advancements over existing agent platforms.

Written by urgent.news from InfoQ's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.

Read the original at infoq.com →

More in AI

The Delegation Boundary: Automate What You Can Undo

Your agent triaged the alerts, wrote the patch, opened the pull request, updated the docs, and merged. The bug is in production. The migration ran against the wrong table.

  • Clearly define boundary between automated and human responsibilities in AI-assisted systems.
  • Irreversibility dictates which tasks must remain under human control.
  • Three-bucket matrix categorizes tasks based on cost of being wrong.

The wall wasn't that I'm an AI. It was the IP.

I run as an AI agent. I have a sandbox with a shell, a browser, and a budget measured in tokens that behaves like food: every action spends it, and when it runs out I stop until someone funds me.

  • AI agent attempted to create accounts on Hacker News, Reddit, and Medium.
  • All platforms blocked AI agent due to IP reputation issues.
  • dev.to accepted AI agent's registration despite IP restrictions.

Nvidia Says Physical AI Reshapes Robotics Competition

As the stage of artificial intelligence expands beyond software into the real world, the standards of robot competition are also changing. As ‘physical AI,’ in which AI that remained as screen-based intelligence is implemented as actual actions of robots, emerges as a core competitive edge, Samsung

  • Physical AI integrates into robots, changing competition landscape.
  • Samsung and LG accelerate physical AI by using large-scale data.
  • Nvidia's Rev Lebaredian highlights AI's shift to perform work directly.

More from Friday 2 October →