Everything OpenAI Announced at DevDay 2026

OpenAI used its 29 September DevDay to launch GPT-6.1 Sol at a fifth of GPT-6 Astra's token prices, computer use in the Agents API, Codex in the cloud and a $500-a-month Pro 500 plan with a faster Ultrafast tier. They were among more than 20 announcements listed in OpenAI's recap.

This round-up groups the launches by product, with the plans each one needs, as OpenAI's own posts, help center and developer docs listed them on 30 September.

The Headline Launch: Always-On Agents

The keynote's lead launch was dots, always-on agents in ChatGPT that run on GPT-6 Astra with their own cloud computer, for Pro and Business Premium users. Our report on OpenAI's dots covers how they work, what they cost, their safeguards, and how they compare with Muse, Grok Bot and OpenClaw.

Models: GPT-6.1 Sol, Ultrafast, and Pro 500

GPT-6.1 Sol is an upgrade to GPT-6 Sol that OpenAI says nearly matches GPT-6 Astra on agentic coding, computer use and professional work at one-fifth of Astra's standard token prices. In the API it is gpt-6.1-sol, and it keeps GPT-6 Sol's input and output prices while halving cached input.

Model	Input	Cached	Output
gpt-6.1-sol	$2	$0.10	$10
gpt-6-sol (22 Sep launch)	$2	$0.20	$10
gpt-6-astra	$10	$1	$50
gpt-6-astra Ultrafast	$60	$6	$300

Prices are US dollars per million tokens for prompts up to 272K input tokens, from OpenAI's API pricing page and API changelog, read 30 September. The gpt-6-sol row is its launch price from the 22 September changelog entry, and the pricing page's main table now lists gpt-6.1-sol in its place.

OpenAI's own benchmarks put GPT-6.1 Sol level with Astra on DeepSWE v1.1 at about one-fifth of the cost. On Terminal-Bench Science 0.1 it cost $5.47 per task against $23.80 for Astra and $23.21 for Claude Opus 5.5, though Astra kept the top score at 68.1%. In ChatGPT, GPT-6.1 Sol is available in ChatGPT Work and Codex on Plus, Pro, Business, Enterprise and Edu, and OpenAI says it is not yet in regular Chat. Our GPT-6 Sol and Luna report covers the model it replaces.

Ultrafast is a premium speed tier, and OpenAI says it generates tokens up to 8 times faster in Codex, at 300 tokens per second, and up to 6 times faster in the API. GPT-6 Astra Ultrafast is live in the Responses API through service_tier: "ultrafast", with global processing and US data residency only, and in ChatGPT Work and Codex on Pro 500 and Enterprise. OpenAI says GPT-6.1 Sol Ultrafast will follow in the coming days.

Pro 500 is the plan that brings Ultrafast to individuals. OpenAI's Pro tiers help article lists three monthly Pro plans: Pro 100 at $100, Pro 200 at $200, and Pro 500 at $500. OpenAI's recap says Pro 500 carries 25 times the ChatGPT Plus usage allowance, the highest of any plan, and it is the only Pro tier with Ultrafast, which uses the plan's included usage first and then draws on credits.

Pro 200 is open to new subscribers again, at the same $200 but with a lower usage allowance for new subscriptions. Pro 200 subscribers who qualify for grandfathering keep their previous allowance through 29 October 2026, as long as the subscription stays active.

Codex: Cloud, CLI, Review, and Security

• Codex in the cloud: start and manage Codex tasks from any device, including a phone, with reusable development environments. Plus, Pro, Business, Healthcare, Education and Enterprise.
• Refreshed Codex CLI: start and steer tasks by voice, a new /agents view for juggling several tasks, better session resume and worktrees, and a cleaner terminal UI. All plans.
• Code Review: a review view in the ChatGPT desktop app with summaries, diffs and the option to ask Codex questions before you comment on a GitHub pull request or GitLab merge request, plus automatic reviews in the cloud. All plans.
• Codex Security Cloud: scans whole GitHub repositories on demand or on a schedule, checks new commits, deduplicates findings and prepares fixes in the cloud, with Daybreak Blue models included and no separate Daybreak application. Pro, Business, Enterprise and Edu, on desktop and web.

The Codex changelog also lists GPT-6.1 Sol for Codex and ChatGPT Work, and new Codex CLI releases, all dated 29 September.

The API: Agents, Decisions, and Bedrock

The Agents API, in public beta since 10 September, now has computer use. The API changelog says agents complete tasks in an OpenAI-hosted browser, while your application handles website access approvals and sign-in. OpenAI's recap adds Codex multi-agent, tool search, tool calling and context compaction, with OpenAI running the infrastructure, and says the capability is also in Codex and ChatGPT Work on Pro 500 and Enterprise.

The Decisions API points OpenAI's Luna model at a question you define with a fixed set of answers, from text or image input, for jobs like classifying content, routing requests, or picking an agent's next action. It is in limited preview, and OpenAI plans a broad release in the coming days.

Bedrock Managed Agents, powered by OpenAI, built with Amazon, takes the core of the Agents API and runs OpenAI agents entirely inside AWS, connected to AWS resources.

ChatGPT for Teams

• ChatGPT Space: a team home where teammates, ChatGPT and each person's dot share knowledge. Pro, Business and Enterprise on desktop and web; mobile can read and share now, with creation later.
• Pages: a new document type that people and agents write together, with research, charts and images. Pro, Business and Enterprise.
• Collaborative slides: coming in the next few weeks, with export to PowerPoint or Google Slides. Pro, Business and Enterprise.
• Teams and shared tasks: team tasks that run on a schedule or when an event fires. Business and Enterprise.
• @ChatGPT in Slack and Microsoft Teams: teammates can mention ChatGPT without an individual license. Business and Enterprise.
• Meetings plugin: notes and summaries saved to Space, with the audio deleted once notes are ready. Beta in the macOS desktop app for Pro and Business, with Enterprise to follow.

Plugins and MCP Events

• Plugin extensions let developers build inside ChatGPT: a sidebar home, interactive panels and viewers for their own file types. All plans.
• Plugin Creator, a redesigned submission flow and better ranking and discovery for plugins. All plans.
• Sites can host plugins on Business, Enterprise, Healthcare and Edu.
• MCP Events: ChatGPT supports the proposed MCP Events specification, so a plugin can start an automation when something happens in a connected app, even while you are away. All plans.
• Shareable profiles display the Sites and plugins someone has built, on Free, Go, Plus, Pro, Business and Enterprise.

Plans and Partners

Sign in with ChatGPT lets people use their ChatGPT plan allowance in 16 partner tools, including Cognition's Devin, Notion, Vercel, T3, OpenClaw and Dactyl, with a control for how much each one can use. The sign-in itself is available globally, and the plan-usage part is for Plus and Pro users in participating tools.

For companies, the OpenAI Marketplace lets eligible enterprises apply part of an existing OpenAI commitment to approved partner software, starting with 32 partners that include Adobe, Figma, Salesforce, ServiceNow, Palo Alto Networks and CrowdStrike. For now, enterprise customers can register their interest with OpenAI. OpenAI Private Intelligence offers Zero Data Retention with Private Safety Processing, which runs automated safety reviews without OpenAI staff seeing the content, and a Private Inference preview using confidential computing is due this fall, through sales.

What Matters for Developers

Most of the list is gated by plan, so the useful first question is which tier you pay for. Here is what each tier gains from DevDay, per OpenAI's recap and launch posts:

Plan	New From DevDay
All plans, Free included	Codex CLI refresh, Code Review, plugin extensions, MCP Events
Plus	GPT-6.1 Sol in Work and Codex, Codex cloud, partner sign-in usage
Pro 100 and Pro 200	Plus items, plus dots, Space, Pages, Codex Security Cloud
Pro 500 ($500/month)	Pro items, plus Ultrafast, and Agents API computer use in Codex and ChatGPT Work
Business	GPT-6.1 Sol, Codex cloud and Security, Space, Pages, team tasks, @ChatGPT, Sites plugins
Business Premium	Business items, plus a dot
Enterprise	Business items, plus dots beta (admin enables), Ultrafast and computer use in Codex and Work, Marketplace interest
Edu, Healthcare	Dots beta (admin enables), Codex cloud, Sites plugins; Edu also gets GPT-6.1 Sol and Codex Security Cloud

The launches that cost nothing extra are the ones marked for all plans. The refreshed Codex CLI with its /agents view and the Code Review view in the desktop app work on any plan, and plugin authors can read the proposed MCP Events specification against their own integrations now.

For API work, the direct saving is GPT-6.1 Sol. It lists at one-fifth of Astra's input and output prices and matched Astra on OpenAI's DeepSWE run, so a workload running on gpt-6-astra can be re-run on gpt-6.1-sol and compared on your own evals before switching. Teams building agents can test computer use in the Agents API beta, while the Decisions API needs a limited-preview slot for now.

Ultrafast is the one launch priced as a premium. In the API it costs six times Astra's standard rate per token, and in ChatGPT it needs Pro 500 or Enterprise. OpenAI positions it for workloads where speed matters most.

What Is Still to Come

DevDay itself was a single day at Fort Mason in San Francisco, and the event page says sessions beyond the livestreamed keynote are recorded and posted on openai.com after the event. OpenAI is also taking DevDay Exchanges to Bengaluru, Tokyo, Seoul, Paris, Berlin, London, São Paulo and Mexico City later this year, without dates on the page.

Several of the launches still have open dates on OpenAI's pages:

• GPT-6.1 Sol Ultrafast and the Decisions API's broad release, both promised within days.
• Collaborative slides in the coming weeks, and the Meetings plugin for Enterprise.
• Private Inference "this fall", with no date or price.

CodeWalkers had not tested the new products when this was written, so every capability here comes from OpenAI's own pages, read in a browser on 30 September 2026. This article will be updated with an editor's note as those launches land.