OpenAI Unveils Dots Agentic AI and GPT-6.1 Sol at Dev Day
Capabilities and Deployment of Dots Agents
Powered by GPT-6 Astra, Dots agents feature their own cloud computer and plug into over 4,000 applications based on human-set boundaries. Altman stated during his keynote that the tools can work continuously to manage bug reports, draft pull requests, or handle budget planning cycles.
Independent technology analyst Carmi Levy noted that Dots represents OpenAI’s definitive entry into AI agents, giving enterprise decision-makers a familiar adoption path built on pre-existing investments in the company’s stack. Dots is currently available to ChatGPT Pro, Business Premium, and Enterprise users, with specialist enterprise versions slated for integration into Microsoft Agent 365.
Pricing and Performance of GPT-6.1 Sol
Alongside Dots, OpenAI launched GPT-6.1 Sol as a faster and more affordable sibling to GPT-6.1 Astra, which was shelved earlier due to safety concerns during internal testing. Sol offers near-Astra level intelligence at one-fifth of the price, with cached inputs priced at $0.10 per million tokens.
According to Altman, Sol provides up to 8x faster token generation in an upcoming ‘Ultrafast’ plan and serves as a daily workhorse for multi-step business workflows and code debugging. Levy observed that the lower cost curve removes a major financial barrier for enterprises deploying AI agents at scale.
Cloud Codex and Governance Features
OpenAI also expanded its developer tooling with Codex Cloud, allowing coding tasks to move seamlessly between devices without losing active threads. The new Codex CLI incorporates voice control for managing tasks locally, while Codex Security Cloud aids defenders in identifying vulnerabilities.
Addressing governance during the keynote, Altman emphasized that the technology should function as a collaborative renaissance rather than an industrial replacement for human workers. Data published by OpenAI shows that Codex adoption has expanded rapidly across non-technical departments such as legal and recruiting, accounting for over 85 percent of output tokens for internal workers.