OpenAI Finally Ships GPT-5.6 to Everyone, Pairs It With New 'ChatGPT Work' Agent

OpenAI's GPT-5.6 is no longer stuck in regulatory limbo. Roughly two weeks after the model launched only to government-approved organizations under a "limited preview," OpenAI has secured the Trump administration's approval to release it broadly. Sam Altman, the company's CEO, called it "the best model we have ever produced."

The company didn't just flip a switch on the model, either. Alongside the wider rollout, OpenAI introduced a new AI agent called ChatGPT Work, built to run on the GPT-5.6 model suite, which includes three models: Sol, Terra, and Luna.

ChatGPT Work is essentially a mashup of ChatGPT and Codex, OpenAI's coding-focused tool. The idea is to hand non-technical users the kind of capability that's historically been reserved for developers. According to OpenAI's blog post, the agent "can gather context from the apps, files, and workflows you choose and create finished materials such as documents, spreadsheets, presentations, and web apps." A new "unified plugins directory" lets it plug into everyday tools like Slack, Gmail, Google Drive, calendars, and CRM software, so it can pull information from where people actually work rather than starting from a blank prompt.

Access is rolling out unevenly, at least at first. Mac and Windows users everywhere, including people on the free tier of ChatGPT, get immediate access to both ChatGPT Work and GPT-5.6 through the desktop app. On mobile and the web, Pro, Enterprise, and Edu subscribers are first in line, with Plus and Business users following within a few days. OpenAI says the global rollout will continue "gradually toward full availability over the next 24 hours."

Why the delay happened

The backstory here matters. GPT-5.6's original debut was unusually constrained: it launched only to organizations that had cleared government approval, a limited preview that kept the model out of ordinary users' hands while officials reviewed it. That kind of gated rollout is not the norm for a flagship AI model release, and it left OpenAI in a holding pattern until Washington signed off. Now that the green light has arrived, the company is treating the public launch as a moment worth pairing with a splashy new product rather than a quiet model swap.

The agent wars are heating up

ChatGPT Work lands squarely in the middle of an increasingly crowded fight to build an AI agent that ordinary people actually want to use for daily tasks, not just coding or research. OpenAI and Anthropic have been racing each other, but they're not alone. Google and, more recently, Apple have all been pushing into the same territory, all trying to claim ground that opened up further after the open-source agent OpenClaw went viral and reset expectations for what a consumer AI assistant should be able to do.

Results across the industry have been mixed. Despite the flurry of releases, nobody has yet delivered the kind of dependable, do-it-all AI agent that can act as an everyday right hand for the average person. That gap is exactly what OpenAI is trying to close with ChatGPT Work, which sits as a direct rival to Anthropic's Claude Cowork, itself a fusion of Claude and Claude Code aimed at the same non-technical audience.

Sol takes center stage

Of the three models powering GPT-5.6, OpenAI is putting the most weight behind Sol, which it's positioning as setting "a new standard for intelligence and efficiency." The company is emphasizing Sol's strength in coding, cybersecurity, science, and computer-use tasks, the kind of work that requires a model to actually operate software rather than just answer questions about it.

OpenAI is also leaning into price as a selling point, marketing Sol as a cheaper option than rival companies' top-tier models. That pitch lands at a moment when the AI industry is under growing pressure over costs. Complaints about a broader money squeeze across AI labs, and about those costs getting passed down to customers, have been building for a while, and OpenAI seems intent on using GPT-5.6's pricing as a point of contrast against competitors.

With the public rollout underway and ChatGPT Work pushing into new use cases, the real test now is adoption. Whether everyday users actually reach for an AI agent to build spreadsheets and presentations, rather than just chatting with a bot, will say a lot about whether this round of the agent race produces a genuine shift in how people work or just another crowded feature launch.