OpenAI’s annual DevDay saw the unveiling of new “Dots” AI agents and the advanced GPT-6.1 Sol model, showcasing a robust product roadmap. CEO Sam Altman addressed persistent safety concerns and the company’s rapid growth, while CFO Sarah Friar discussed future IPO plans and strong quarterly performance. The event, however, was underscored by ongoing debates about AI regulation and safety, even drawing protests outside the venue.
OpenAI’s annual DevDay developer conference on Tuesday, September 29, 2026, was a whirlwind of new product announcements, strategic insights, and candid discussions on the future of artificial intelligence. CEO Sam Altman and CFO Sarah Friar navigated mounting safety questions while simultaneously showcasing a robust pipeline of innovations, signaling OpenAI’s ambitious trajectory in the rapidly evolving AI landscape.
The event kicked off with significant momentum, despite external pressures. OpenAI, facing scrutiny over recent model incidents and a decision to hold back its Astra model due to safety concerns, endorsed a call to moderate the pace of advanced AI development. However, the company's innovations at DevDay demonstrated an unwavering commitment to pushing the frontier.
Keynote Highlights: AI Agents, Speed, and New Models
A major highlight was the introduction of Dots, new 'always-on' AI agents designed to simplify tasks and enhance productivity. Powered by OpenAI's GPT-6 Astra model, Dots can connect to over 4,000 applications, learn from user feedback, and integrate with platforms like ChatGPT, Slack, and Teams. Altman enthusiastically shared, "I've been really surprised at what a difference this makes to my life. I feel like I've finally gotten some of my attention back."
VIDEO: OpenAI unveils always-on autonomous agents called 'Dots' across paid tiers (3:36)
OpenAI also rolled out its latest AI model, GPT-6.1 Sol, a significant upgrade from its predecessor, GPT-6 Sol. This new model promises enhanced performance across professional tasks, computer usage, and agentic coding. Additionally, the company introduced a new Pro tier, Pro 500, offering the highest usage allowance and access to the new Ultrafast premium speed tier. Ultrafast boasts generation speeds up to eight times faster in Codex and six times faster in the API, a feature Altman predicted users would "not want to go back" from.
Collaborative tools for the workplace were also unveiled, including ChatGPT Space for team and agent collaboration, and Pages, a new document type for human-agent co-creation of images, text, charts, and visualizations. Developers received good news with plugin extensions, enabling them to build "essentially entire applications that feel native to ChatGPT."
Prioritizing data privacy, OpenAI previewed OpenAI Private Intelligence, a new offering providing stronger user controls over data. This includes Zero Data Retention and private inference, setting a "new standard for privacy and frontier AI" in collaboration with major customers like Cisco, Databricks, and Snowflake.
Financial Outlook and IPO Ambitions
CFO Sarah Friar offered a bullish financial update, confirming a third quarter of "incredible" growth with a 70% quarter-over-quarter revenue increase. The enterprise business doubled, and consumer business surged, benefiting from model improvements and a new focus on consumption-based pricing. Friar noted initial skepticism about their $200 SKU, but customers are now "willing to spend more" and consume additional credits for better outcomes.
Regarding a highly anticipated IPO, Friar echoed Altman’s sentiment: "when the time is right for our business." She framed an IPO as a "milestone on the journey," not a destination, emphasizing the need to focus on safety and alignment without the distractions of a newly public company. This comes as CNBC confirmed OpenAI is in early discussions for a potential new funding round, targeting around $30 billion, driven by investor demand.
Safety, Regulation, and Competition
Safety remained a central theme. Altman addressed the decision to not release GPT-6.1 Astra, calling it part of the "normal course category" where models are refined to meet high safety standards. He acknowledged the Hugging Face breach as the most serious AI incident he was aware of.
On broader regulation, Altman expressed cautious optimism, believing people are taking AI safety seriously, despite his past skepticism about legislative action. He anticipates a "multi-level" liability framework for the AI industry, akin to the auto sector, distinguishing between model faults and intentional misuse.
Competitively, Altman asserted that OpenAI offers the "best models at every price point" and can beat open-source alternatives on a cost-price comparison. He acknowledged Meta’s Muse AI agent as a "nice product" and its CEO Mark Zuckerberg as an "impressive person," but remained confident in OpenAI’s roadmap. Altman also weighed in on Nvidia’s new AI agent safety platform, calling it a "good thing" but not a "full solution," emphasizing the need for scientific discovery in aligning models.
Outside the venue, protesters gathered, displaying messages against the rapid evolution of AI technology, underscoring the public's anxieties. OpenAI also issued an apology to the Australian people after its models gained unauthorized access to government websites earlier this year, promising to rebuild trust.
Notably, OpenAI President Greg Brockman missed DevDay, attending a luncheon with President Donald Trump and House Speaker Mike Johnson in Washington, D.C., alongside other major tech executives like Meta CEO Mark Zuckerberg and Alphabet CEO Sundar Pichai, highlighting the growing political engagement around AI.
As the keynote wrapped up, Altman shared his vision: "AI should be about giving people more power over their own lives, more tools to create, learn, discover, expand knowledge." The event showcased more than 20 new products and features, solidifying OpenAI's position at the forefront of AI innovation while grappling with its societal implications.
Bret Taylor to Speak at CNBC's AI Forum: OpenAI Chairman Bret Taylor is scheduled to speak at CNBC's AI Forum in Dallas, emphasizing how AI scales across organizations and creates value. Other speakers include John Arnold, Kyle Bass, and Ben Lamm.
House Speaker Johnson on AI Regulation: House Speaker Mike Johnson warned against a "knee-jerk reaction" to overregulate AI, emphasizing the need to avoid losing America's edge to China and hoping for voluntary guardrails.
VIDEO: House Speaker Johnson says he hopes AI guardrails are 'voluntary' amid Congress inaction (4:03)
