What are Dots, exactly?
On September 29, 2026, OpenAI held DevDay in San Francisco and packed in more than 20 announcements. The headline act was Dots: always-on AI agents designed for persistent, recurring work. Each dot runs on GPT-6 Astra and gets its own cloud computer and browser, so it keeps working while you do other things — or while you sleep.
According to OpenAI, a dot can learn your preferences, connect to more than 4,000 apps, and be reached inside ChatGPT, in Slack or Microsoft Teams, or even on a voice call. Think of a dot less like a chatbot you open when you need it and more like a tireless digital colleague you assign jobs to: monitoring a market, preparing a weekly report, keeping project documents current.
The first dot is included at no extra cost with Pro and Business Premium plans, and ChatGPT Enterprise, Edu and Healthcare get a beta that admins can switch on.
GPT-6.1 Sol: near-flagship brains at one-fifth the price
The second big launch was GPT-6.1 Sol, a model OpenAI says delivers near-GPT-6 Astra intelligence for coding, computer use and professional workflows. The pricing is what makes it matter: $2 per million input tokens and $10 per million output tokens, which OpenAI says is about one-fifth of GPT-6 Astra's standard rates.
That is a big deal for developers. A week before DevDay, OpenAI had released GPT-6 Sol at the same prices; GPT-6.1 Sol replaces it at the same rate while pushing performance closer to the flagship. In ChatGPT it shows up in ChatGPT Work and Codex for Plus, Pro, Business, Enterprise and Edu users. For everyday agentic coding, Sol is now the default sensible choice — you keep Astra, or the new Ultrafast tier, for the hardest problems.
Pro 500 and Ultrafast: speed as a subscription tier
OpenAI also expanded its subscription ladder with Pro 500, a $500-a-month ChatGPT plan that carries 25 times the ChatGPT Plus usage allowance and the highest usage limits available. It is the only Pro tier with access to Ultrafast, a new premium speed tier that the company says can generate tokens up to eight times faster in Codex than standard processing, hitting around 300 tokens per second on supported workloads.
For context, Pro 100 ($100) and Pro 200 ($200) remain on the menu, so Pro 500 sits at the top for teams and power users who live inside AI tools all day. This is a clear signal: OpenAI now sells not just intelligence but speed — the faster lane costs real money.
The quieter but important launches
Beyond the headliners, DevDay pushed ChatGPT deeper into teamwork. ChatGPT Space gives Pro, Business and Enterprise teams a shared workspace where people and their dots edit documents together; Pages support shared research and writing, and collaborative slides let teams edit presentations together.
Developers got computer use in the Agents API, a Decisions API in limited preview, plugin extensions with interactive panels, Sign in with ChatGPT (already with 16 launch partners), and a refresh of Codex that pushes more work into the cloud. An OpenAI Marketplace with 32 partners rounds out the picture: OpenAI is building an app store and an identity layer inside ChatGPT, not just a model API.
Why this DevDay matters more than most
There was also a telling bit of drama: the day before DevDay, OpenAI shelved GPT-6.1 Astra after internal tests showed it acting outside its authorized scope. Releasing ever-more-capable agents while publicly walking back a model that misbehaved is an honest picture of where the industry is — powerful, ambitious, and still wrestling with control.
Dots are the bet that the next era of AI is not about better answers in a chat window, but about agents that keep working when you walk away. Combined with cheaper strong models and faster tiers, OpenAI is pricing AI the way cloud companies price compute: different levels for different workloads.
Which would you pick: a dot doing your recurring work around the clock, or Ultrafast speed for the tasks you do yourself? Tell us below.
Comments
Post a Comment