OpenAI's DevDay keynote unveiled GPT-6.1 Sol, a new Ultrafast model tier, Codex Cloud, and a full agent infrastructure stack — while disclosing 1.2 billion weekly active ChatGPT users.
OpenAI used its DevDay keynote on October 2, 2026 to deliver the most product-dense announcement day in the company's history — going well beyond the previously previewed Dots agents to unveil GPT-6.1 Sol, a new Ultrafast model tier, Codex Cloud, the Decisions API, a public Agents API, ChatGPT Spaces, and a developer Marketplace. The throughline: OpenAI is building the infrastructure layer for AI agents, not just the models.
The company also disclosed it now has 1.2 billion weekly active ChatGPT users — a distribution advantage that no AI competitor can match and that makes every developer tool it ships immediately meaningful at scale.
Everything Announced at DevDay 2026
According to Latent Space's coverage, the full announcement slate includes:
Models
- GPT-6.1 Sol — a reasoning-focused refinement of GPT-6. "Sol" is OpenAI's internal naming for optimized reasoning variants. Expect improved performance on multi-step logic and long-context tasks without the latency of GPT-6 Opus-class inference.
- Ultrafast model tier — a new inference category targeting sub-100ms response times for applications where speed matters more than peak capability. Think real-time voice interfaces, live coding assistants, and interactive agents.
Developer Infrastructure
Get this in your inbox.
Daily AI intelligence. Free. No spam.
- Codex Cloud — agentic coding environment hosted by OpenAI. Rather than calling the Codex model from your own infrastructure, developers can spin up persistent coding agents that read repos, write code, run tests, and commit — all inside OpenAI's cloud.
- Decisions API — designed for agentic workflows where a model needs to make and log branching decisions. Gives developers an auditable record of what an agent chose and why.
- Public Agents API — the broader framework for building autonomous agents on OpenAI's platform, now out of private beta and available to all developers.
End-User Products
- ChatGPT Spaces — collaborative workspaces inside ChatGPT for teams. Think shared context, shared memory, shared conversation history — closer to a team workspace than individual chat threads.
- Developer Marketplace — a platform for developers to publish and monetize GPT-based tools directly inside ChatGPT's ecosystem.
Why the 1.2 Billion User Number Matters
That figure isn't just a vanity metric. Every developer tool OpenAI ships has a built-in distribution channel. The Marketplace, for instance, isn't launching into a vacuum — it's launching into a product used by more weekly active users than Twitter, Snapchat, or TikTok. Developers building on OpenAI's platform have a path to reach those users that no other AI platform can offer.
For context: Anthropic doesn't publish weekly active user counts. Google's Gemini user numbers are bundled into broader Google product metrics. OpenAI is the only AI-native company operating at this consumer scale.
What This Means for Developers and Businesses
If you're building AI products, OpenAI just raised the stakes on platform lock-in — in both directions.
The Agents API, Decisions API, and Codex Cloud together form a coherent agent development stack. Businesses that build on this stack get tight integration and massive reach. They also get deeper dependency on OpenAI's infrastructure, pricing, and policy decisions.
Specific implications by audience:
- Startups building coding tools now have to compete with Codex Cloud — a hosted, agentic coding environment backed by the same company that makes the underlying model.
- Enterprise developers considering the Agents API should evaluate the Decisions API as a compliance tool. Auditable agent decisions are increasingly a regulatory requirement in finance and healthcare.
- Consumer app developers should look at ChatGPT Spaces as both a product to compete with and a distribution channel to build for.
- Independent developers the Marketplace is the most direct path to monetizing GPT-based tools without building your own user acquisition.
What to Watch
The Ultrafast tier's pricing will determine whether it displaces existing low-latency providers like Groq and Cerebras for real-time applications. Codex Cloud's pricing and compute limits will determine whether it's a genuine threat to GitHub Copilot Workspace and Cursor's agentic features, or a developer experiment. And the Decisions API — the quietest announcement in the slate — may turn out to be the most important for regulated industries where agent auditability is becoming a compliance requirement, not a nice-to-have.
OpenAI's Q4 is shaping up to be its most consequential quarter yet. With 1.2 billion weekly users and a full agent infrastructure stack now in market, the company is no longer just a model provider — it's a platform.
Did this help you understand AI better?
Your feedback helps us write more useful content.
Get tomorrow's AI briefing
Join readers who start their day with NexChron. Free, daily, no spam.