OpenAI released GPT-5.6 Sol, its most powerful artificial intelligence model, on Thursday alongside a new workplace tool called ChatGPT Work. The launch, which follows a delay prompted by U.S. government cybersecurity concerns, intensifies the race for AI dominance with rival Anthropic.
A model that matches the competition
GPT-5.6 Sol roughly matches the performance of Anthropic's leading model, Fable 5, on standard benchmark tests, according to Vals AI, a firm that tracks AI systems. Rayan Krishnan, chief executive of Vals AI, said that in financial and legal tasks, the model is highly capable. "In real-world situations involving financial and legal tasks, it is state of the art," he said.
ChatGPT Work, powered by Sol, is an AI agent that can operate software applications and websites for users. It handles spreadsheets, online calendars, and email services, similar to Anthropic's Claude Cowork. The tool signals a broader push toward agents that perform multi-step office tasks without constant human guidance. ChatGPT Courses cover the underlying platform and its evolution, including the agent capabilities now central to ChatGPT Work.
Government oversight slows the rollout
The release took shape after the U.S. government moved to restrict both OpenAI's and Anthropic's newest models over cybersecurity risks. Last month, President Trump signed an executive order asking technology companies to voluntarily give the government oversight of new AI models before public release. The order marked a shift from the administration's earlier hands-off approach to AI regulation.
The delay did not prevent the launch, but it added a layer of scrutiny that is likely to persist as models grow more powerful. OpenAI's decision to proceed reflects the competitive pressure to keep pace with Anthropic, which has made rapid advances with its own models and workplace agents. OpenAI Courses explore the company's technology trajectory, including the model architectures and safety practices that inform releases like Sol.
Why this matters for IT, development, and research professionals
For technical and research teams, GPT-5.6 Sol delivers performance that rivals the current state of the art in financial and legal reasoning. The agent capabilities in ChatGPT Work point to a near future where AI tools integrate directly with everyday software stacks, automating tasks across spreadsheets, calendars, and email. Professionals who understand these models and their limitations can make sharper decisions about where to deploy them and how to manage the associated security and compliance risks.
Your membership also unlocks:
