OpenAI’s July 9 rollout is more than another model announcement. GPT-5.6 is now available across ChatGPT, Codex, and the OpenAI API, and ChatGPT Work adds a more hands-on layer for longer tasks, connected tools, and finished deliverables. For business teams, that combination matters because it shifts the conversation from “Can the model answer?” to “Can it finish the workflow?”
The new release is aimed squarely at professional work. OpenAI says GPT-5.6 comes in three tiers—Sol, Terra, and Luna—and that ChatGPT Work can gather context, plan the approach, and take action across tools, files, and desktop apps to create spreadsheets, docs, slides, and Sites.
What OpenAI actually launched
On the model side, GPT-5.6 is shipping as a family instead of a single flagship. OpenAI positions Sol as the most capable tier, Terra as the balanced option, and Luna as the fastest and most affordable option. On the product side, ChatGPT Work is available on desktop today and is rolling out to web and mobile on paid plans over the next few days.
OpenAI also added more explicit workflow controls. In ChatGPT Work, the ultra setting is available to Pro and Enterprise users, while the API adds multi-agent support in beta so GPT-5.6 can run concurrent subagents and synthesize the results in one request. That is a meaningful signal: OpenAI is designing for structured execution, not just single-turn replies.
Why this matters for business teams
The practical change is not that AI got smarter in a vacuum. It is that the product now assumes work happens across documents, files, apps, and recurring tasks. That is closer to how operations, finance, marketing, sales, and engineering teams actually work.
For buyers, that means two things. First, the bar for adoption is higher: a useful system must fit permissions, review steps, and data access without creating chaos. Second, the upside is higher too: if the workflow can be decomposed into context gathering, drafting, checking, and delivery, a model like GPT-5.6 can do more than answer questions—it can move work forward.
What to watch before you roll it out
Teams should treat this launch as a prompt to review their automation map. The right question is not whether to use GPT-5.6 everywhere. The right question is which workflows are ready for delegated execution, which still need human approval, and which should be handled by a dedicated agent or team instead of a general-purpose assistant.
- Use it where the work is document-heavy, repeatable, and reviewable.
- Keep human checks in the loop for sensitive, public, or high-stakes outputs.
- Test whether desktop access, connected files, and recurring tasks actually reduce friction.
- Separate one-off assistance from workflows that deserve a dedicated AI worker or coordinated team.
That is where the Nerova lens matters most: launches like this are strongest when they help teams decide what should become a chatbot, a single agent, or a multi-step AI team.