Skip to content
Vibe Code Basics
AI News

OpenAI DevDay 2026 Recap: Dots, GPT-6.1 Sol, Codex Cloud, Agents API & More

Everything OpenAI announced at DevDay 2026 and what it means for vibe coders: Dots, GPT-6.1 Sol, Codex Cloud, Code Review for GitHub and GitLab, the Agents API, Ultrafast mode, the $500 Pro tier, and Sign in with ChatGPT.

By Vibe Code Basics Editorial TeamPublished 4 min read
On this page (11)

OpenAI's DevDay on September 29, 2026 was its biggest developer event yet. The headline was Dots, always-on personal agents powered by GPT-6 Astra. But for people who build software, the bigger story may be everything around it: GPT-6.1 Sol with near-flagship quality at a fifth of the price, Codex Cloud for tasks that keep running across devices, Code Review for GitHub and GitLab, a public beta of the Agents API, a new Ultrafast mode, and a $500/month Pro tier.

Here's every announcement that matters to vibe coders, in plain English. For the full month of AI news, see our October 2026 roundup.

The announcements at a glance

AnnouncementWhat it isWho gets it
DotsAlways-on agents with their own cloud computerChatGPT Pro, Business Premium; Enterprise beta
GPT-6.1 SolImproved Sol at $2/$10 per million tokensAPI, Codex
Codex CloudCodex tasks that keep running across devicesPlus and above
Code ReviewAutomated PR review for GitHub (GA) and GitLab (preview)Codex users
Codex Security CloudSecurity scanning for codeLaunched same day
Agents APIManaged agent sessions, multi-agent, computer use, memoryPublic beta
UltrafastMuch faster generation at a premium pricePro 500, Enterprise
Pro 500$500/month plan with Astra UltrafastNew plan
Sign in with ChatGPTLet users log in to your app with ChatGPT16 launch partners

Sources: BenchLM's DevDay summary (opens in a new tab) and TechCrunch on Dots (opens in a new tab).

Dots

Personal agents with a cartoon avatar, a cloud computer and a browser, connected to thousands of apps, which keep working in the background and can hand coding work to Codex. We cover them in depth in OpenAI Dots explained.

GPT-6.1 Sol

OpenAI pitched GPT-6.1 Sol as "near-Astra capability at a fifth of the price", keeping Sol's $2 input / $10 output pricing. For most vibe coding in Codex, it's the model you'll actually use day to day. Compare it with Luna and the flagship in GPT-6 Sol vs GPT-6 Luna and GPT-6 Astra explained.

Codex Cloud

Codex tasks can now run in OpenAI's cloud and continue across devices: start a task on your laptop, check on it from your phone, review the result on your desktop. Cloud tasks get the same tools as local Codex, including plugins and computer use. It's available on ChatGPT Plus, Pro, Business, Education, Healthcare and Enterprise.

Why it matters: this is the same direction as Claude Code on the web, Cursor's cloud agents and Copilot's coding agent. Agents increasingly work while you don't, which makes clear specs and good review habits essential. See spec-driven development.

Code Review for GitHub and GitLab

Codex's automated pull request review is now generally available for GitHub, with GitLab support in preview. It joins Copilot code review, Cursor's Bugbot and Claude's review tools. Having a second AI review every PR is one of the cheapest quality wins available to a solo vibe coder.

Codex Security Cloud

Launched alongside DevDay, Codex Security Cloud focuses on finding vulnerabilities in your code. Given the wave of coding agent security issues disclosed around the same time, it's good timing. It complements, rather than replaces, a pre-launch security checklist.

Agents API (public beta)

For developers building their own agents, the Agents API provides managed sessions, multi-agent controls, computer use and memory, so you don't have to build that infrastructure yourself. OpenAI also previewed a Luna-powered Decisions API for fast, cheap selection from predefined options.

Why it matters: if you're vibe coding an AI-powered app, these APIs reduce how much agent plumbing you (or your AI) need to write. Keep the usual rules in mind: server-side API keys, per-user limits and spending alerts.

Ultrafast and the $500 Pro plan

Ultrafast delivers much faster generation (OpenAI said up to 8x in the keynote; API docs say up to 6x) at six times the standard price. It's available for Astra on the new Pro 500 plan ($500/month, about 25x Plus usage) and Enterprise.

For most vibe coders, this is overkill. It's aimed at people running agents all day, where waiting is the bottleneck.

Sign in with ChatGPT, plugins and the marketplace

  • Sign in with ChatGPT lets users log into third-party apps with their ChatGPT account (16 launch partners).
  • Plugin extensions from partners like Figma and Adobe Photoshop extend what ChatGPT and agents can do.
  • OpenAI Marketplace lets enterprises buy partner tools using existing OpenAI commitments.

What it means for vibe coders

  1. Cheaper good-enough AI. GPT-6.1 Sol makes high-quality coding assistance cheaper; check whether your plan or API spend can come down.
  2. Background agents are the new normal. Codex Cloud and Dots push work into the background. Invest in clear task descriptions and review.
  3. Free code review. If you're on a ChatGPT plan with Codex, turn on Code Review for your repositories.
  4. Pick tools by workflow, not hype. OpenAI's stack is now a strong alternative to Claude Code and Cursor. Compare them in our AI coding tools guide, and see which model to use in best AI model for coding.

Frequently asked questions

When was OpenAI DevDay 2026?

September 29, 2026.

What was the biggest announcement at DevDay 2026?

Dots, OpenAI's always-on personal agents. For developers, GPT-6.1 Sol and Codex Cloud are arguably just as important.

Is Codex Cloud free?

It's included with paid ChatGPT plans from Plus upward, subject to usage limits.

What is the OpenAI Agents API?

A public-beta API for building agents, with managed sessions, multi-agent controls, computer use and memory built in.

  • #OpenAI
  • #DevDay
  • #Codex
  • #OpenAI Dots
  • #GPT-6
  • #AI News

Keep reading