OpenAI launches ChatGPT Work productivity agent with GPT-5.6, computer control, and task scheduling
OpenAI has released ChatGPT Work, a productivity agent powered by its new GPT-5.6 models that combines ChatGPT, Codex coding agent, and web browsing capabilities in a single application. The agent includes computer control features, remote task scheduling, and unified plugin integration accessible through desktop apps on Mac and Windows, plus web access for paid subscribers.
OpenAI launches ChatGPT Work productivity agent with GPT-5.6, computer control, and task scheduling
OpenAI has released ChatGPT Work, a unified productivity agent that consolidates ChatGPT, its Codex coding agent, and web browsing capabilities into a single application. The agent is powered by OpenAI's new GPT-5.6 model family, which was previously delayed by the Trump administration.
ChatGPT Work is accessible through a redesigned ChatGPT desktop app available on Mac and Windows, with Plus and Pro subscribers also gaining access through the web client. Free tier users can access the agent exclusively through the desktop application.
Key capabilities
The agent includes computer control features that allow it to operate across applications and move files on a user's system. Task scheduling enables users to assign work from mobile devices, with the agent executing tasks remotely while users track progress through desktop or web interfaces.
A unified plugin directory uses an "@" mention system for connecting to external apps and services. Users can type "@" followed by an app name in prompts to specify which services ChatGPT should access or use for context. The system also provides automatic app recommendations during conversations.
According to OpenAI, the task scheduling feature can automate routine work. The company cites Slack message summarization as an example use case, where ChatGPT can compile weekly message summaries automatically.
Browser integration and Sites feature
The redesigned desktop app includes a built-in browser that replaces OpenAI's standalone Atlas browser, which the company plans to discontinue on August 9. James Sun, who worked on Atlas, confirmed the deprecation timeline.
A new feature called Sites allows users to generate web applications directly through ChatGPT. OpenAI states Sites are designed for "live dashboards, project trackers, launch calendars, prototypes, internal portals, and interactive reports."
Privacy controls
OpenAI emphasizes user control over the agent's capabilities. Users determine which applications and data ChatGPT can access, set check-in frequencies, and specify which actions require manual approval before execution.
Rollout details
OpenAI began rolling out ChatGPT Work today and expects to complete deployment within 24 hours. Pricing for the underlying GPT-5.6 models has not been disclosed.
What this means
ChatGPT Work represents OpenAI's most significant move into enterprise productivity software, directly competing with Microsoft Copilot and Google's Workspace AI tools. The computer control features and task scheduling suggest OpenAI is positioning the agent as a replacement for robotic process automation tools, though the actual reliability and scope of autonomous task execution remain to be tested in production environments. The integration of coding, browsing, and productivity features into a single agent follows the industry trend toward unified AI assistants, but success will depend on whether the agent can reliably handle complex, multi-step workflows without excessive user intervention.
Related Articles
Simon Willison's Pelican Benchmark Shows GPT-6 Astra Outperforming GPT-5.6 Sol at Every Reasoning Level
Developer Simon Willison ran his signature 'pelican riding a bicycle' SVG test on newly-accessed GPT-6 Astra across five reasoning levels, comparing results against GPT-5.6 Sol, Terra, and Luna. Even Astra's lowest reasoning setting reportedly beat every Sol output, though Astra costs roughly twice as much per token.
OpenAI Lists GPT-6 Astra Pro on OpenRouter: Same Model, Higher-Compute Reasoning Mode
GPT-6 Astra Pro, now listed on OpenRouter, is the existing GPT-6 Astra model configured to run with reasoning.mode set to 'pro' for higher-quality output on complex tasks. It carries a 1M-token context window and tiered pricing from $5/$25 to $20/$100 per million input/output tokens depending on the serving tier.
OpenAI's GPT-6 Astra Cuts Hallucinations, But Indirect Prompt Injection Attacks Still Succeed 8.5% of the Time
OpenAI's new GPT-6 Astra model shows major improvements in hallucination rates and jailbreak resistance over predecessor GPT-5.6 Sol, according to OpenAI's system card. However, indirect prompt injection attacks hidden in documents still succeed 8.5% of the time in external testing by Gray Swan, down from 27% but still above rival Claude Opus 5's 4.8% rate.
OpenAI's GPT-6 Astra Splits Benchmarks but Beats Human Efficiency on ARC-AGI-3, Pushing Chollet's AGI Timeline Forward
OpenAI's GPT-6 Astra rates first place on Epoch AI's aggregate benchmark but ties its predecessor on Artificial Analysis. Its human-beating move efficiency on ARC-AGI-3 led ARC Prize co-founder François Chollet to call progress 2x faster than expected.
Comments
Loading...