By ByteBot
OpenAI held DevDay 2026 on September 29 in San Francisco and announced Dots — always-on AI agents powered by GPT-6 Astra that keep working after you close the laptop. Each dot gets its own dedicated cloud computer, its own browser, and access to over 4,000 connected apps. The framing from OpenAI: stop babysitting your AI and start delegating to it.
How Dots Actually Work
Each dot runs on GPT-6 Astra and operates in two modes. In idle mode, a dot performs read-only proactive research — it can gather information and monitor for changes, but it cannot send messages, modify app content, or take any action. When you are active or have explicitly granted permission, a dot shifts into full computer-use mode: it can write code, run tests, submit pull requests, and interact with connected apps.
The safety architecture matters before you hand over credentials. Dots ship with built-in rules governing autonomous versus approval-required actions. Users can configure Custom Rules — allow, block, or require approval for specific work categories. Sensitive actions like password changes always require explicit user confirmation. A monitoring system can pause or stop a dot if anomalous behavior is detected. You can open the dot’s cloud computer at any time to inspect exactly what it is doing.
For developers, OpenAI’s stated use case is instructive: a dot can monitor incoming customer feedback, identify recurring bug patterns, scope the fixes, build and test them, then return completed pull requests with video walkthroughs — without you touching anything between assignment and review.
GPT-6.1 Sol Is the Real Story for Developers
Dots grabbed headlines, but GPT-6.1 Sol may matter more to developers building agentic pipelines. OpenAI launched Sol alongside Dots, positioning it as near-Astra performance at one-fifth the cost.
Pricing: $2 per million input tokens, $10 per million output tokens, and $0.10 per million for cached input — a 95% discount on prompts reusing the same context. Independent analysis puts Sol at roughly 22% of Astra’s per-task cost for equivalent work. On DeepSWE v1.1, Sol matches GPT-6 Astra while beating the previous GPT-6 Sol by 6.4 percentage points. On AutomationBench, Sol scores 2.2 points above Opus 5.5 at a third of the cost. The 95% cache discount changes the economics of long-context agentic loops — pipelines that reuse system prompts and shared context could see costs fall sharply.
Dots Enters a Market Meta Already Owns
Dots is not launching into a vacuum. SpaceXAI shipped Grok Bot in beta on August 11. Meta followed with Muse on September 8 — and within two weeks, Muse had overtaken ChatGPT as the top free iOS app in the United States, collecting more downloads than both Claude and Grok.
OpenAI’s play is differentiated. Dots connects to 4,000+ apps versus Muse’s tighter integration list. More significantly, OpenAI is targeting enterprise with specialist dots — agents that carry their own organizational credentials for roles like procurement, invoice processing, and customer support. Microsoft is integrating specialist dots into Agent 365. On consumer, OpenAI is playing catch-up. On enterprise, it is staking a claim before Muse gets there.
Other Developer Announcements Worth Knowing
DevDay dropped 20+ announcements. The developer-relevant ones beyond Dots and Sol:
- Codex Security Cloud — scans your repositories and prepares security fixes automatically.
- Code Review — analyzes GitHub PRs and GitLab MRs on an ongoing basis.
- Agents API update — now supports multi-agent loops with computer use; AWS Bedrock Managed Agents integration included.
- Decisions API — handles classification and routing tasks with predefined answer sets. Limited preview now, broad release within days.
- Plugin extensions — build full apps that run natively inside ChatGPT.
- Sign In with ChatGPT — 16 launch partners including Notion, Vercel, and Devin let your existing Pro or Plus plan cover API usage in their apps.
Pricing Changes — Including One Downgrade
A new Pro 500 plan launches at $500/month with 25x Plus allowance and Ultrafast access — up to 300 tokens per second at 6x standard pricing. Dots are included with existing Pro and Business Premium plans at no extra cost.
The change fewer outlets are flagging: the Pro 200 plan ($200/month) now offers 10x Plus for new subscribers, down from the previous 20x. Existing subscribers keep their current terms, but the value proposition has shifted for anyone evaluating the plan.
What to Do With All of This
The always-on agent race moved fast — three major entrants in under two months. For developers, the immediate priority is GPT-6.1 Sol’s pricing, not Dots’ existence. The 95% cache discount is the kind of number that makes previously uneconomical architectures viable. Dots is the product vision. Sol is the infrastructure that makes building toward it affordable. Full DevDay announcement details are here.
