AI

OpenAI DevDay 2026: GPT-6.1 Sol, Always-On Dots Agents and More

OpenAI used DevDay 2026 to launch GPT-6.1 Sol at about a fifth of GPT-6 Astra's price, plus always-on "Dots" agents. Here is what builders should know.

KIING DIGITAL LABS · · 4 min read

On Tuesday, Sept. 29, OpenAI held DevDay 2026, its annual event for developers, and the headline was clear: more capable AI is getting much cheaper. The company introduced a new model called GPT-6.1 Sol, a set of always-on AI agents called "Dots," and a long list of tools for people who build software on top of its platform. OpenAI also said ChatGPT now has 1.2 billion weekly users.

What happened

The biggest launch for developers was GPT-6.1 Sol. According to OpenAI and reporting from The Next Web, the model is priced at $2 per million input tokens and $10 per million output tokens through the API, which works out to roughly one-fifth of the price of GPT-6 Astra, OpenAI's top model. OpenAI described Sol as offering "near-Astra intelligence for a fifth of the price," and said it nearly matches Astra on coding and computer use. The Next Web reported that Sol is available in Codex, in the API and in ChatGPT's work features for paying plans.

The second headline was Dots, which OpenAI describes as always-on agents. Unlike a normal chatbot that waits for you to type, a Dot runs in the background inside ChatGPT, works toward goals you set, and can connect to more than 4,000 apps through plugins, according to BGR. OpenAI shared an example of an early tester whose Dot noticed he had forgotten to invoice a publication, prepared the invoice and sent it after he approved it. By default, Dots need the user's approval before sending messages, editing content or controlling a computer, and at launch they are limited to ChatGPT's Pro 200 and Business Premium plans.

OpenAI also announced:

  • Ultrafast, a premium speed option for GPT-6 Astra that OpenAI says generates tokens up to eight times faster in Codex.
  • Codex in the cloud, which lets developers run coding tasks in remote environments and check on them from other devices, including a phone.
  • ChatGPT Space, a shared workspace where teams, ChatGPT and Dots can work on projects together.
  • A Decisions API, built to help companies automate narrow, repeatable choices such as classification and routing, launching first as a limited preview.

Why it matters

Price is the story. When a model that performs close to the top of the market costs about 80 percent less, a lot of product ideas that did not make financial sense last month suddenly do. Features like AI-written reports, automated customer replies or coding assistants inside an app all depend on the cost of each request. Lower costs mean better margins for software companies and lower prices for their customers.

Dots matter for a different reason. They show where the big AI companies think software is heading: away from tools you open and click through, and toward agents that do the work for you in the background and connect to all the apps you already use. That puts pressure on every app to be "agent-friendly," meaning it can be reached and used by an AI assistant, not just by a person.

It also raises the bar for what customers expect. If ChatGPT can remember to send an invoice, users will start asking why their other software cannot.

What this means for founders and builders

If you are planning a software product right now, three practical lessons stand out.

First, recheck your math. If you priced an AI feature using last quarter's model costs, the numbers may have changed in your favor. Cheaper models make it realistic to offer AI features on lower-priced plans, or even in a free tier.

Second, build for agents as well as people. Products that expose clean integrations, clear data and simple actions will be easier for assistants like Dots to use. That can become a new source of customers.

Third, speed to market matters more than ever. The platform is moving every few months. Founders who spend a year building from scratch risk launching into a market that has already shifted. One way to move faster is to start with a product that is already built and working. At KIING DIGITAL LABS, we build complete software products, including SaaS and AI tools, that are ready to launch. You can browse products ready to acquire or, if you need something specific, start a custom build designed around today's AI models from day one.

The bottom line

OpenAI's DevDay 2026 delivered two big signals: frontier-level AI is getting dramatically cheaper, and the next wave of software will be built around agents that work in the background. For founders, that means more room to build profitable AI features, and less time to waste getting to market.

Free tool: Have an idea sparked by this news? Turn it into a product plan in minutes with Concept Studio — free. Skip the build: Explore fully built products ready to acquire or start a custom build.

  • OPENAI
  • AI MODELS
  • AI AGENTS
  • DEVELOPER TOOLS

From the lab

KIING DIGITAL LABSAI

AI

Meta Releases Muse Wonder, Its First Frontier AI Model

Meta Superintelligence Labs launched Muse Wonder and Muse Wonder Full, two large models built for long-running AI agents that remember what matters to users.

October 13, 2026 · 4 min readRead article