OpenAI launches ChatGPT Work, artificial intelligence for automating complex tasks
OpenAI announced the global availability of ChatGPT Work on Friday, July 10, 2026. It is a new artificial intelligence agent designed to perform complex actions across multiple applications and files, manage long-running projects, and turn ambitious goals into concrete results. This innovation is driven by Frontier’s latest family of AI models, GPT-5.6, which includes Sol, Terra and Luna variants.
An AI assistant for multifaceted workplace tasks
ChatGPT Work expands the capabilities of artificial intelligence by acting as an autonomous agent. It can gather contextual information from user-selected applications, files, and workflows, using this data to create ready-made materials such as documents, spreadsheets, presentations, and even web applications. This functionality promises to optimize professional routines, allowing AI to deal with activities that require multiple steps and coordination.
The tool represents an advance in the integration of AI with the daily and complex demands of the corporate environment. The ability to operate across different platforms and data formats eliminates the need for constant human intervention in repetitive or time-consuming tasks, freeing professionals to focus on more strategic and creative activities.
On the same topic: OpenAI introduces new GPT-5.6 family and ChatGPT Work agent for complex tasks
The force behind ChatGPT Work: the GPT-5.6 family
Behind ChatGPT Work’s versatility is OpenAI’s new line of GPT-5.6 templates, which includes three variants optimized for different needs. The flagship model, Sol, sets a new standard in intelligence and efficiency, demonstrating superior performance in areas such as programming, knowledge-based work, cybersecurity and science, while consuming fewer tokens and having a reduced cost.
Completing the family, the Terra is a balanced model, ideal for daily work, offering a robust combination of performance and practicality. Luna was developed with a focus on cost-benefit, making cutting-edge artificial intelligence more accessible. This diversity allows users to choose the model that best suits the complexity and budget of their tasks.
Superior performance and efficiency: the results of the new models
The GPT-5.6 family not only promises, but delivers impressive results in several performance evaluations. The new models significantly outperform their predecessors and competitors in key benchmarks, setting new standards for artificial intelligence in professional and technical environments.
Among the main statistics that demonstrate progress, the following stand out:
See Also
Michigan serial killings result in six deaths, including US suspect
7.7 magnitude earthquake destroys buildings in Indonesia and triggers tsunami warning
Videos of the 7.7 earthquake in Indonesia reveal damage and panic in Asia
More on this story: End of ChatGPT Atlas: OpenAI redirects navigation with AI for apps and extension
- Agents’ Last Exam: The GPT-5.6 Sol achieved a record 53.6 points, surpassing the Claude Fable 5 (adaptive) by 13.1 points. At the average reasoning level, Sol outperformed Fable 5 by 11.4 points, at approximately a quarter of the estimated API cost. The Terra and Luna models also surpassed the Fable 5, at about one-sixteenth the cost.
- Artificial Analysis Coding Agent Index: The GPT-5.6 Sol established a new level of excellence with 80.0 points, surpassing the Claude Fable 5 by 2.8 points. This performance was achieved with less than half the output tokens, in less than half the time and costing about a third less.
- ExploitBench (Cybersecurity): An assessment measuring the ability to progress from accessing vulnerable code to arbitrary code execution showed GPT-5.6 Sol achieving 73.5%, a notable improvement compared to GPT-5.5’s 47.9%, while maintaining a comparable output token budget.
- Scheduling Agent Tasks: GPT-5.6 Sol, with maximum reasoning, outperformed Claude Fable 5 using 54% fewer output tokens. This advantage extends across the entire family: GPT-5.6 Terra, in its highest performance configurations, outperformed Claude Fable 5, and GPT-5.6 Luna outperformed Opus 4.8, both at approximately one-fifth the estimated API cost and one-third the time.
Access and availability: who can use the new tool
The availability of ChatGPT Work and GPT-5.6 models is happening gradually globally. The ChatGPT desktop application, which integrates Chat, Work and Codex functionalities into a single interface, is now accessible to Mac and Windows users, including the Free category.
For Pro, Enterprise and Edu users, ChatGPT Work will begin rolling out on the web and mobile devices, with access extended to the Plus and Business categories in the coming days. The GPT-5.6 Sol and Pro models will also be available for different subscription levels, allowing Pro and Enterprise users to select the GPT-5.6 Pro for the most demanding tasks.
How ChatGPT Work will impact everyday professional life
The introduction of ChatGPT Work promises to transform the way professionals and companies deal with the volume and complexity of their daily tasks. By automating long-running workflows and coordinating multiple steps, the tool frees teams to focus on innovation and strategic decisions, rather than getting lost in repetitive processes.
Learn more: OpenAI expands ad testing on ChatGPT to boost revenue
Integration with a unified plugin directory allows ChatGPT Work to connect with essential tools like Slack, Gmail, Google Drive, calendars and CRMs, creating a more cohesive and efficient work ecosystem. The addition of a built-in browser and the ability to operate your computer expands the possibilities even further, making AI a partner capable of doing deeper work on local websites and applications. With scheduled tasks, the tool ensures that recurring activities are kept moving without constant supervision.
