OpenAI DevDay 2026: Dots agents, GPT-6.1 Sol and Pro 500, and ChatGPT becoming the app store itself

The headline item is Dots. Per the official description, Dots are remarkably capable, always-on agents built to handle everything: they get to know what matters to you, keep working on your behalf, and take important work off your plate. Availability has explicit edges. Pro and Business Premium users get them in eligible markets; Enterprise, Edu and Healthcare users can try the beta only after a workspace admin enables it, and it is off by default. TechCrunch adds a key detail: each Dot has its own cloud computer and browser, can use connected apps, and can reach an ecosystem that now spans more than 4,000 apps. In other words, a Dot is not answering questions in a chat box; it is completing work on a machine of its own.

The other half is the models. OpenAI describes GPT-6.1 Sol as a major upgrade to GPT-6 Sol with exceptionally strong performance on agentic coding, computer use and professional work, delivering near-Astra intelligence at a fifth of standard input and output token prices so developers can run more complex workflows within the same budget. On speed there is Ultrafast, which OpenAI calls its premium speed tier for workloads where speed matters most: up to 300 tokens per second in Codex, roughly 8 times faster generation, and up to 6 times in the API. Availability is stated precisely. GPT-6 Astra Ultrafast is available today in the API and in ChatGPT Work and Codex on Pro 500 and Enterprise plans; GPT-6.1 Sol Ultrafast is coming soon. Mashable reports the new Pro 500 tier costs 500 dollars a month with 25 times the ChatGPT Plus allowance and includes Ultrafast; the official post states only the multiple and the inclusion, so the price figure is Mashable's.

What actually changes distribution is identity and allowance. OpenAI says Sign in with ChatGPT lets you use your ChatGPT plan with other tools, sharing allowance across 16 partners including Cognition's Devin, Notion, Vercel, T3, OpenClaw and Dactyl, with controls over how much each can use; eligible usage counts toward your plan limits. Identity is available globally and plan-usage sharing is available to Plus and Pro users in participating tools. TechCrunch nails why this layer matters: as more discovery happens inside AI chat, carrying a user's existing AI allowance with them is an incentive a distribution channel cannot offer. On the enterprise side there is the OpenAI Marketplace, where eligible customers can apply part of their existing OpenAI commitment toward approved partner software. OpenAI lists 32 launch partners spanning Adobe and Figma for creative work, Sierra, Decagon, HubSpot, Salesforce and ServiceNow for customer experience, Harvey and Legora for legal, Palo Alto Networks and CrowdStrike for security, and Baseten for open-source models.

On developer tooling, the official post lists several concrete changes. Codex can now run wherever you need it: on a computer, remotely from a phone, or in the cloud from any device, with reusable development environments that speed up task starts and give teams a shared setup with approved settings and permissions. The Codex CLI got a refresh. For enterprise data protection there is Private Intelligence, where Zero Data Retention with Private Safety Processing enables automated safety reviews without giving OpenAI personnel access to the underlying content; a preview of Private Inference combining confidential computing with strict, verifiable controls arrives this fall.

TechCrunch's reporting goes deeper into the experience layer, and it deserves attention. ChatGPT will suggest an app inside the flow of conversation when it recognizes the app could help, and the user can connect it and work with it directly inside ChatGPT. The plugin architecture expands with extensions, letting developers build interactive panels so users operate tools while chatting; things that previously required a web page or a native app now happen inside the conversation. Plugin submission also changed: developers can track review, see what needs fixing, request a human review, and update tools without restarting the whole submission. Dots' proactivity has limits: they perform what OpenAI calls proactive research using connected apps in a read-only mode, but users keep control over the actions taken. One gap must be stated plainly: TechCrunch notes OpenAI announced no billing or revenue-sharing system comparable to the economic layer of traditional app stores.

One note on how to read a launch like this. The Verge had previewed that OpenAI teased more than 20 launches and quoted Altman saying we have found a new thing; CNBC and Mashable each ran live and roundup coverage. Reading the official post against the media reports, two kinds of things actually landed. One is capability that can be priced today, such as GPT-6.1 Sol's price points and Ultrafast's token throughput. The other is structure that takes time to show its weight, such as the identity layer, the allowance layer and the review process. The second kind does not pay off on launch day, but it decides whether developers move their apps in. Judging whether this approach works is not about counting announcements; it is about how many apps genuinely put their main experience inside ChatGPT a few months from now, and when the missing billing plan shows up.