After its competitors: OpenAI unveils its own agent | Everything you need to know about it

OpenAI unveiled at DevDay 2026 the Dots platform – autonomous AI agents operating in the cloud, alongside the dramatically discounted GPT-6.1 Sol model and the ChatGPT Space workspace. CEO Sam Altman presented a direct response to competitors, with agents that learn the user and perform complex tasks around the clock.

OpenAI held its annual developer event DevDay 2026 on Tuesday, with a series of significant announcements. CEO Sam Altman presented the Dots platform – autonomous AI agents powered by the GPT-6 Astra model, running on an isolated cloud computer and capable of performing complex tasks around the clock. Each user can assign a dot a name, identity, and goals, and it learns the user's preferences and conducts proactive research even when inactive. The system only contacts the user for decisions or approval of critical steps. The competitive advantage lies in connecting to over 4,000 applications through the plugin ecosystem, including Slack and Microsoft Teams. Organizations are offered Specialist Dots with access permissions to internal systems. Concurrently, OpenAI launched the GPT-6.1 Sol model, delivering performance close to the Astra supermodel on complex programming tasks at a fifth of its cost – $2 per million input tokens and $10 per output. Cache memory call prices were slashed by 95% to 10 cents. The model matched Astra's performance on the DeepSWE v1.1 benchmark, surpassed Claude Opus 5.5 at a third of the cost on AutomationBench, and reduced factual error rates by 32%. OpenAI also launched ChatGPT Space, a collaborative workspace with interactive Pages documents that update in real time, directly competing with Google Workspace. Additionally, a Pro 500 subscription tier was introduced at $500 per month with ultra-fast access, and the retirement of the GPT-5.5 model was announced for mid-next month. To allay security concerns, OpenAI presented the Private Intelligence concept with a zero-data-retention mechanism and private security processing.