OpenAI launches GPT-5.6 and ChatGPT Work AI agent for automated tasks

OpenAI has unveiled GPT-5.6 and a new AI agent called ChatGPT Work, marking a significant expansion of its capabilities in enterprise automation and workflow management. The release follows a period of regulatory scrutiny, with the U.S. Government initially restricting the model’s public rollout over national security concerns. After implementing additional safety measures, the administration approved the launch, which OpenAI described as a milestone in AI-driven productivity.

The GPT-5.6 model family includes three variants: Sol, Terra, and Luna. Sol, the most powerful, is designed for complex tasks requiring advanced reasoning, while Terra balances performance and cost for everyday work. Luna prioritizes speed and efficiency for high-volume operations. OpenAI positioned the models as a response to competitive pressures, particularly from Anthropic’s Claude Cowork, which it aims to outpace in terms of price, speed, and enterprise integration.

ChatGPT Work, the AI agent, represents a rebranding and expansion of OpenAI’s Codex tool, now integrated into a unified ChatGPT desktop app. The agent can autonomously handle multi-step workflows, such as generating documents, analyzing data, and coordinating with external applications like Slack, Microsoft Teams, and Google Drive. It operates across web, mobile, and desktop platforms, with Pro, Enterprise, and Edu users gaining immediate access. Plus and Business tiers will follow in the coming days.

A key feature of ChatGPT Work is its “Unified Plugins Directory,” which aggregates third-party integrations and allows users to trigger specific tools via commands like “@appname.” The agent can also execute scheduled tasks, such as compiling weekly reports or updating presentations based on new data. OpenAI emphasized its ability to “gather context across connected apps and files,” reducing repetitive administrative work. However, the company acknowledged the need for user oversight, particularly in high-stakes scenarios.

The U.S. Government’s involvement in the rollout highlights growing tensions between AI innovation and regulation. The Trump administration’s initial delay was tied to concerns about GPT-5.6’s coding and cybersecurity capabilities, which critics feared could be exploited for malicious purposes. OpenAI addressed these worries by implementing “safety guardrails” and establishing a direct communication channel with federal agencies. The company also noted a 40% reduction in “hallucinations” compared to previous models, alongside a context window of 2 million tokens for processing extensive datasets.

Analysts and early testers offered mixed assessments. While some praised GPT-5.6’s reliability and cost-efficiency, others argued that Anthropic’s Fable model still outperformed it in raw intelligence. OpenAI’s CEO, Sam Altman, highlighted the model’s 54% improvement in token efficiency for agentic coding tasks, framing it as a tool for enterprises seeking to optimize spending. Meanwhile, critics raised concerns about the concentration of AI power within a single company and its implications for labor markets, with some predicting a shift toward “algorithmic management” in corporations.

The release also signals OpenAI’s broader strategy to merge its tools into a “super app” that spans coding, research, and business operations. By integrating Codex into ChatGPT and expanding features like “Computer Use” and “Sites,” the company aims to streamline workflows for both developers and non-technical users. However, the complexity of the new interface has drawn comparisons to a “video game difficulty screen,” with some users finding the array of models and modes overwhelming.

As OpenAI moves forward, the success of GPT-5.6 and ChatGPT Work will depend on enterprise adoption, regulatory compliance, and the ability to balance automation with human oversight. The models’ impact on industries ranging from software development to corporate management remains to be seen, but their launch underscores the accelerating pace of AI’s integration into professional workflows.