OpenAI has received Trump administration clearance to ship GPT-5.6 to the general public, ending a roughly two-week limited preview period during which the model was available only to government-approved organizations. The company paired the broader rollout with a new agent product called ChatGPT Work, which combines ChatGPT with Codex and is powered by the GPT-5.6 model suite. Global availability is set to complete within 24 hours, with Mac and Windows desktop users — including free ChatGPT users — getting access immediately.
Sam Altman framed GPT-5.6 as the strongest system OpenAI has shipped to date. The suite comprises three models: Sol, Terra, and Luna. Sol sits at the top of the stack and is being marketed for coding, cybersecurity, science, and computer-use tasks, with OpenAI positioning it as a lower-cost alternative to competitors' flagship models at a moment when AI labs are passing rising compute costs to customers.
“the best model we have ever produced.”— Sam Altman, OpenAI CEO
ChatGPT Work is the more consequential product launch of the day. It extends Codex-style capabilities — previously the domain of engineers — to non-technical users, letting them generate documents, spreadsheets, presentations, and small web apps from the same interface used for chat. The pitch is that a marketer or operations lead can now use the same agentic tooling that a developer would.
Key facts
- 01GPT-5.6 exits a two-week limited preview restricted to government-approved organizations and begins a global public rollout within 24 hours.
- 02The GPT-5.6 suite consists of three models: Sol, Terra, and Luna, with Sol positioned as the top-tier model for coding, cybersecurity, and science.
- 03ChatGPT Work combines ChatGPT with Codex capabilities and connects to Slack, Gmail, Google Drive, calendars, and CRMs via a unified plugins directory.
- 04Mac and Windows desktop users, including free ChatGPT users, get immediate access; Plus and Business web/mobile users follow over the next few days.
- 05The product competes directly with Anthropic's Claude Cowork, which pairs Claude with Claude Code.
The connective tissue is a unified plugins directory that lets ChatGPT read from Slack, Gmail, Google Drive, calendars, and CRM systems. That plugin surface is what turns ChatGPT Work from a smarter chatbot into a workflow tool: the agent can pull context from where work already happens rather than requiring users to paste material into a prompt.
“It can gather context from the apps, files, and workflows you choose and create finished materials such as documents, spreadsheets, presentations, and web apps.”— OpenAI, company blog post
Access rolls out in tiers. Desktop users on Mac and Windows, including on the free plan, have immediate access via the ChatGPT desktop app. On mobile and the web, Pro, Enterprise, and Edu subscribers get access first, with Plus and Business subscribers following over the next few days. OpenAI said the rollout will reach full global availability within 24 hours.
The competitive frame is direct. ChatGPT Work is squarely aimed at Anthropic's Claude Cowork, a product that similarly bundles Claude with Claude Code to bring agent capabilities to a general audience. Google and Apple have also been building toward the same target — the everyday consumer's right-hand AI agent — with mixed results so far. The viral open-source project OpenClaw raised the bar in recent months and pushed the incumbents to move faster.
OpenAI's cost positioning matters here. The company is explicit that Sol is meant to undercut competitors' most powerful models on price while matching them on capability. That messaging arrives against a backdrop of industry-wide margin pressure — inference costs remain the central variable in AI unit economics, and every frontier lab is looking for ways to hold the line on end-user pricing without eroding gross margin.
The regulatory backstory is worth noting. GPT-5.6 spent roughly two weeks in a preview restricted to government-approved organizations before the Trump administration cleared it for public release. That structure — a government-vetted preview preceding a consumer launch — is a departure from prior GPT releases and hints at a more formalized review process for frontier models going forward, at least for OpenAI.
There are open questions. OpenAI has not published benchmark numbers for Sol against Anthropic's Claude Opus line or xAI's recently released Grok 4.5, which shipped at a fraction of Opus 4.7's price. The company's claim that Sol sets a new standard for intelligence and efficiency will need to survive independent evaluation. Agentic reliability — whether ChatGPT Work actually completes multi-step workflows without derailing — is also unproven at scale.
The strategic read is that OpenAI is trying to convert its lead in raw consumer distribution into a defensible position in agentic productivity before Anthropic, Google, or Apple can catch up. ChatGPT already has the largest installed base of any AI product; bolting a plugin directory and Codex-grade execution onto that base is a bet that distribution plus capability beats capability alone. If Sol delivers on the price-to-performance claim and the plugin ecosystem fills out quickly, this becomes the reference product for AI-at-work through 2026. If it lands as another impressive demo that stumbles on real workflows, Claude Cowork gets an opening it did not have yesterday.
Working on something we should cover, or seeing a story we missed? Send leads, documents, or feedback to hello@aichatdaily.com. For sensitive tips, see our secure tips page for Signal and PGP options.
Spotted an error? Email hello@aichatdaily.com with the URL and the issue, or read our full corrections policy.
