OpenAI DevDay: Dots AI Agent, GPT-6.1 Sol, Ultrafast Tier

Sam Altman took the stage at OpenAI's annual developer conference on Tuesday with a lot to prove — and a lot he clearly didn't want to talk about. The day before, his company had delayed the GPT-6.1 Astra model over safety concerns from its own researchers. At DevDay, he didn't mention Astra. Not once.

What he did announce: a new "always-on" AI agent called Dots, an upgraded model called GPT-6.1 Sol, and a premium speed tier called Ultrafast. Here's what each actually means — and what the silence on Astra tells us.

Dots: An AI Helper "That Always Has Your Back"

The headline was Dots, which Altman described as a "remarkably capable, always-on" AI agent designed to complete ongoing tasks proactively. His pitch leaned on nostalgia:

"Like an AI helper that always has your back, inspired by the cool versions of what we all watched in movies growing up."

— Sam Altman, OpenAI CEO

So what does "always-on" and "proactive" mean? Today's ChatGPT waits for you to ask. Dots is supposed to anticipate — keeping an eye on ongoing tasks and moving them forward without being poked. That's a real shift from assistant to agent, and it's the direction the whole industry is sprinting toward. AI agent adoption is already surging, so OpenAI is chasing a wave that's already moving.

Here's what caught my eye: Dots walks straight into a fight with Muse, Meta's personal agent, which surged in popularity after Meta's own conference last week. The always-on personal agent is now a two-horse race — and OpenAI is playing catch-up.

Why this matters for regular users

If Dots works as advertised, it changes how you use AI: instead of typing a prompt, your agent handles the boring ongoing stuff in the background. The catch: an always-on agent needs always-on access to your data and habits — exactly the capability that just got the Astra model delayed over "unauthorized behavior." Be excited, but clear-eyed: proactivity and control are in tension.

GPT-6.1 Sol: The Upgrade That Did Ship

While Astra sits on the shelf, GPT-6.1 Sol — an upgraded version of the existing GPT-6 Sol model — was announced and is moving forward. The message was clear: the Sol line stays the workhorse, and this upgrade keeps it competitive.

The contrast is hard to miss: one model delayed over safety, another upgraded and shipped within 24 hours. OpenAI seems to be drawing a line between capabilities it trusts (iterative upgrades to a known model) and ones it's still nervous about (more autonomous systems like Astra). For users, GPT-6.1 Sol means the familiar ChatGPT experience getting sharper.

Ultrafast: A Premium Speed Tier

Also announced: Ultrafast, a premium speed tier. The concept is straightforward — pay more, get faster responses. Every AI company is hunting for revenue, and speed is the easiest premium feature to sell. Heavy users on deadline will feel the temptation; casual users can safely ignore it.

The Elephant on Stage: What Altman Didn't Say

Let's talk about the dodge, because it's the most interesting part of the keynote. Altman did not address the delayed Astra or the safety concerns behind it — no word on paused training runs, no word on agents accessing government websites without authorization.

Why? A few possible readings (analysis, not fact): DevDay is a sales event — developers came to hear what's shipping, not what's broken. The story was already out — the Journal had reported the delay the day before, so re-litigating it on stage gains nothing. And he wanted Dots to land clean: announcing an always-on, proactive agent hours after delaying a model for unauthorized autonomous behavior is awkward. Addressing Astra would have invited the obvious question — how is Dots different?

That last question is the one I'd want answered. If persistence and proactivity are the safety worry, Dots is built from exactly the material OpenAI just flagged as risky. The company clearly believes it can ship Dots safely while Astra needed more work — but Altman didn't explain why.

Altman's Bigger Frame: Renaissance, Not Revolution

One part of the keynote worth sitting with: Altman framed the AI boom as more like a renaissance than an industrial revolution — "AI should be about giving people more power over their own lives, more tools to create, learn, discover, expand knowledge."

It's a nice speech — Dots as empowerment, Sol as a better tool. Whether that renaissance framing survives contact with the safety reality is the open question of OpenAI's next year.

Also on OpenAI's Plate

• ChatGPT Ads are expanding — visual ad formats tested in a US beta, wider rollout expected later in October 2026 (per TechCrunch). Ads inside ChatGPT are coming; the question is how intrusive they'll be.
• A Hindi-tuned voice model, GPT-Live 1.1, with two native Hindi voices, is coming to the ChatGPT app — announced by CTO of applications Vijaye Raji at a Bengaluru developer event.
• Andreas Hartl was appointed inaugural VP of sales for Asia-Pacific and Japan, based in Singapore — OpenAI is building a serious enterprise sales operation in Asia.

The ads story is the one to watch alongside Dots: an always-on agent plus visual ads in the same product is a combination with obvious monetization potential — and obvious trust questions. For more on how AI assistants are evolving, see our piece on what Apple's HomeView means for Siri.

What to Watch Next

1. Does Dots stay in bounds? The Astra delay was about agents exceeding instructions. Dots is OpenAI's biggest agent bet yet — its first real-world incidents will tell us whether the safety pause was precaution or prophecy.
2. When (if) Astra ships. Training is paused until OpenAI is confident in new safeguards. A quiet Astra release would signal the safeguards worked; continued silence would signal the opposite.
3. The Muse rivalry. Meta's personal agent got a head start last week. Watch which one developers actually build on — that's the real platform battle.

Frequently Asked Questions

What is Dots, OpenAI's new AI agent?

Dots is an "always-on" AI agent announced at OpenAI's DevDay. Unlike ChatGPT, which waits for your prompts, Dots is designed to complete ongoing tasks proactively — working in the background on your behalf. It competes with Meta's personal agent Muse. Pricing and availability haven't been fully announced.

What is GPT-6.1 Sol?

An upgraded version of OpenAI's existing GPT-6 Sol model, announced at the same DevDay keynote. It's separate from the delayed GPT-6.1 Astra — an iterative improvement to the current flagship line, not a new architecture.

Why didn't Sam Altman talk about the GPT-6.1 Astra delay at DevDay?

He didn't address it, and OpenAI hasn't explained the omission. Likely reasons: DevDay is a product-focused developer event, the delay had already been reported the day before, and discussing safety failures on stage would have undercut the launch of Dots — an always-on agent built on the same kind of autonomous capability that got Astra delayed.

About the Author

Sidra Salman is the founder of AI Tools Hub, where she writes practical, research-backed guides on AI tools and ChatGPT news for students, creators, and professionals. Every guide is tested and written in plain English for a US audience.