OpenAI DevDay 2026: Dots, GPT-6.1 Sol and what matters for your business

Picture a marketing lead at a 12-person company on Monday morning. The weekly report is not built, a product launch changed scope over the weekend, and three social posts need rewriting to match. Until now, AI could help with each of those jobs, but only while someone sat at the keyboard asking. OpenAI's DevDay on September 29, 2026 was largely about changing that: agents that keep working in the background, cheaper models to run them, and shared spaces where a team and its AI work on the same files. ¹

OpenAI called it its biggest DevDay yet, with more than 20 announcements across ChatGPT, Codex, its models and new ways of working, and says its products now reach a combined 1.2 billion weekly users. Plenty of the announcements are aimed at developers. This guide sorts out the ones a business owner or marketer should understand, what they cost, and which are available now versus announced for later. ¹

Dots: AI that keeps working when you stop typing

A dot is an always-on agent that learns what matters to you and works toward your goals around the clock. Each dot runs on GPT-6 Astra, has its own cloud computer and browser, and can reach more than 4,000 apps through ChatGPT plugins. You talk to it in ChatGPT on desktop, web or mobile, or in Slack and Microsoft Teams, and you can hop on a voice call with it. It remembers the context across all of those places. ³

The examples OpenAI gives map neatly onto marketing work. For a content creator, a dot finds clip-worthy moments in a new interview transcript, prepares show notes and drafts social posts for approval, then carries your edits across the rest of the material. For a launch lead, it revises the plan, story and creative drafts when the product scope changes. One early tester's dot noticed he had forgotten to invoice a publication, prepared the invoice and sent it after he approved it. ³

The controls are what make that usable in a real business. When you are not working with it, a dot does what OpenAI calls proactive research using connected apps restricted to read-only access, so it cannot send messages or change anything on its own. Built-in rules decide when it acts and when it asks; Custom Rules let you allow, require approval for, or block specific actions; and an Activity View shows what it did in the background. Sensitive tasks such as changing a password always stay with you, and OpenAI still tells users to review consequential work. ³

Dots are rolling out to ChatGPT Pro and Business Premium users in eligible markets, with Enterprise, Edu and Healthcare workspaces able to try a beta once an admin switches it on. Your first dot is included in the plan, and chatting with it does not count toward your ChatGPT limits, although tasks it starts in Codex or ChatGPT Work do. OpenAI is also starting enterprise pilots of specialist dots with their own company identity, drawing on its internal tests in areas such as invoice processing, customer support and email marketing. ¹³

GPT-6.1 Sol: near-Astra results at a fifth of the price

GPT-6.1 Sol is an upgrade to GPT-6 Sol, the mid-priced model OpenAI released a week earlier. OpenAI says it nearly matches GPT-6 Astra, its most capable model, on agentic coding, computer use and professional work, at one-fifth of Astra's standard input and output token prices. The list price is the same as GPT-6 Sol's; what changed is how much the model does for that price, plus a cheaper rate for reused context. ²

A token is a small piece of text, roughly three quarters of a word, and API prices are quoted per million of them. Cached input is material the model has already seen in a recent request, such as a long brand guide or product catalog that an agent rereads on every task. At $0.10 per million, that is 95% below the standard input price and half of GPT-6 Sol's cached rate, which matters for exactly the kind of always-on agent work DevDay was about. ²

OpenAI's own results tell a consistent story. On AutomationBench, which tests multi-step business workflows across 47 tools in sales, marketing, operations, support, finance and HR, GPT-6.1 Sol scores 2.2 percentage points above Claude Opus 5.5 at medium effort for roughly a third of the cost, and 4.8 points above GPT-6 Sol. On the OSWorld 2.0 computer-use test it lands within 2.1 points of Astra at roughly a seventh of the cost per task. On a factuality test built from deliberately difficult prompts, responses with at least one factual error fell from 11.4% to 7.7% at low effort. ²

Vendor benchmarks deserve a second opinion, so here is Artificial Analysis, which runs its own tests. On its Intelligence Index, a composite of ten evaluations covering professional tasks, coding, research and knowledge, GPT-6.1 Sol at maximum effort scores 51.8 against 52.7 for GPT-6 Astra, while its cost to complete an index task is $0.72 against $3.26. The near-Astra claim holds on that measure. Claude Opus 5.5 at maximum effort still leads this group at 57.6, but at $5.98 per task. ⁴

Two practical notes follow from the chart. First, effort matters as much as the model: GPT-6.1 Sol at medium effort scores about the same as the old GPT-6 Sol at maximum, for about a fifth of the cost per task. Second, in ChatGPT, GPT-6.1 Sol is available to Plus, Pro, Business, Enterprise and Edu users in ChatGPT Work and Codex, not yet in regular Chat. Developers can use it in the API now as gpt-6.1-sol. ²⁴

Ultrafast: paying for speed

Ultrafast is a premium speed tier for work where waiting is the expensive part. OpenAI says it generates tokens up to eight times faster in Codex, around 300 tokens per second, and up to six times faster in the API. GPT-6 Astra Ultrafast is available now in the API, and in ChatGPT Work and Codex on the Pro 500 and Enterprise plans; GPT-6.1 Sol Ultrafast is due in the coming days. ¹²

For most small businesses, Ultrafast is not the first thing to buy. It earns its cost where someone is waiting in real time: a live demo in front of a client, a developer iterating on a website, or an agent that has to answer a customer before they leave the page. Overnight reports and scheduled jobs gain little from it.

Codex and the developer tools behind your website

Codex is OpenAI's coding agent, and it now runs in the cloud from any device, with reusable development environments so tasks start quickly and a team shares approved settings. The Codex command line can take spoken instructions, code review lets Codex take a first pass on changes automatically, and Codex Security Cloud scans GitHub repositories on demand or on a schedule, then prepares fixes even with your laptop closed. ¹

Business owners rarely use these directly, but they shape the work you buy. If an agency or developer maintains your website, ongoing security scanning and automatic first-pass reviews are reasonable things to ask about. On the API side, the Agents API now supports computer use, so developers can build agents that operate ordinary software; OpenAI-powered agents can run entirely inside Amazon Web Services through Bedrock Managed Agents; and a Decisions API, in limited preview, answers fixed questions for tasks like classifying or routing requests. ¹

Plugins turn ChatGPT into a storefront

The quieter announcement with the biggest marketing implication is plugins. OpenAI is opening the platform it uses to build ChatGPT's own features, so companies can give their plugin a home in the ChatGPT sidebar with interactive panels beside the conversation. Improved ranking and recommendations will surface relevant plugins in the directory and inside conversations, and plugins can now start automations when something happens in a connected app, through the proposed MCP Events standard. ¹

For brands, that looks like the early days of app stores: a way to put a booking flow, product finder or quote calculator where ChatGPT's users already are. It is also another reason to make sure AI tools describe your business accurately, because customers can now meet your brand inside an assistant before they ever reach your website. ¹

ChatGPT becomes a shared workspace

Several announcements move ChatGPT from a personal assistant toward a team tool. ChatGPT Space is a shared home where teammates, ChatGPT and their dot build on common knowledge. Pages are collaborative documents where anyone can write, research, chart or create images, and collaborative slides that export to PowerPoint or Google Slides are coming in the next few weeks. Business and Enterprise teams can share recurring tasks such as weekly updates, and anyone can mention @ChatGPT in Slack or Teams without needing their own license. ¹

The Meetings plugin, in beta on the ChatGPT desktop app for Mac, takes notes and saves summaries with action items to your Space, then can update a project plan or draft a follow-up. OpenAI says the audio is deleted once the notes are ready. As with any recording tool, tell the people in the meeting before you use it. ¹

Who can use what, and when

Plans and pricing changes worth knowing

Sign in with ChatGPT lets Plus and Pro subscribers spend their plan allowance inside 16 partner tools, including Notion, Vercel and Cognition's Devin, with a cap for each. The new Pro 500 plan offers 25 times the Plus usage allowance and includes Ultrafast. For larger companies, the OpenAI Marketplace lets eligible enterprise customers put part of an existing OpenAI commitment toward 32 partners, including HubSpot, Salesforce, Adobe and Figma. And OpenAI Private Intelligence adds zero data retention with private safety processing, with a Private Inference preview due this fall. ¹

What this means for marketing teams

The through-line is delegation with guardrails. Dots, team tasks and plugin automations all let AI act while you are away, and each one depends on the permissions you grant. Start with read-only access and approvals on anything that sends, posts or pays, and expand only when the drafts stop needing edits. That is the same advice OpenAI builds into dots by default. ³