OpenAI is racing to wire its most powerful AI yet into everyday office life, even as CEO Sam Altman publicly warns that demand for the new flagship Sol model could outstrip the company’s infrastructure.

June–early July: Government scrutiny, then a green light

OpenAI’s GPT‑5.6 family — Sol, Terra and Luna — was first debuted in June but held back from a full public launch while the Trump administration reviewed its advanced cybersecurity capabilities.¹ The company describes GPT‑5.6 as its “strongest cybersecurity model yet,” promising frontier performance in defensive tasks such as threat modeling and code review.² A detailed system card presents Sol as the flagship, Terra as a lower‑cost option, and Luna as the fastest, most cost‑efficient tier, all shipped with OpenAI’s “most robust yet” safeguards.³

After receiving government approval in early July, OpenAI rolled GPT‑5.6 out to the public, with Sol positioned as the workhorse model for enterprise work, coding and scientific research.²¹

July 9: Biggest office push yet

On July 9, OpenAI launched ChatGPT Work, an agent that can “stay with a project for hours if needed, and turn a goal into finished work,” integrating across apps, files and workflows.⁴ The same day, GPT‑5.6 became the preferred model inside Microsoft 365 Copilot for Word, Excel, PowerPoint, Chat and Cowork,⁵ with Microsoft’s Satya Nadella praising its “stronger reasoning and higher‑quality outputs without sacrificing efficiency.”Satya Nadella (satyanadella)⁶

Human‑side coverage framed this as OpenAI’s “biggest play for the office,” merging Codex into the ChatGPT desktop app and introducing a Work setting meant to turn ChatGPT into a “super app for work.”⁷ Tech outlets noted that ChatGPT Work can automate end‑to‑end workflows, from analyzing budgets to generating marketing campaigns, and even run scheduled tasks autonomously when users are away.⁸ Ars Technica asked bluntly: “Give ChatGPT access to all your work tools; what could go wrong?”⁸

At the same time, new “hosted sites” features led some to ask whether OpenAI is coming for web‑builder platforms next, arguing that tools aimed at internal decks could easily grow into a full site builder.⁹

Mid‑July: Performance hype meets scale worries

From OpenAI’s own perspective, GPT‑5.6 is quickly becoming a backbone model. The company highlighted that it is now the preferred engine for Microsoft 365 Copilot,⁵ while Altman amplified research claiming “physicians found fewer flaws in GPT‑5.6 responses than physician‑written responses,” positioning it as a major step forward for health intelligence at lower cost.Sam Altman (sama)¹⁰

Third‑party platforms raced to adopt the models as well. TechCrunch reported that GPT‑5.6 is aimed squarely at rivals like Anthropic, with OpenAI calling Sol its “best coding model yet” and emphasizing strong cyber‑defense capabilities.² Perplexity and Amazon Bedrock added GPT‑5.6 across their stacks, with AWS describing three “tiers of intelligence” built for scale and reliability.¹

Yet just days after the fanfare, Altman publicly cautioned that “5.6 sol growth is insane” and that, despite “heroic work” by the inference team, “it is possible there are some hiccups soon.”¹¹Sam Altman (sama)¹² Axios framed this as a reminder that even leading AI firms can struggle to scale infrastructure fast enough when usage spikes, especially amid a broader arms race in frontier models.¹¹

How the perspectives differ

OpenAI’s own messaging stresses capability, safety and ubiquity: GPT‑5.6 is “state of the art at reasoning through multi‑step tasks,” powering ChatGPT Work to “get real work done across web, mobile, and desktop.”⁴ The GPT‑5.6 system card foregrounds tiers and safeguards to argue the models can be deployed “safely and at scale, around the world.”³

Human‑authored coverage is more ambivalent. Business Insider and The Verge see a strategic push to dominate office workflows and the AI‑agent race,⁷¹ while Ars Technica dwells on practical and security risks of letting a long‑running agent roam across Slack, Google Drive, desktops and browsers.⁸ The Verge also underscores the political dimension, noting that GPT‑5.6’s public rollout came only after a “limited preview” restricted to government‑approved organizations.¹

Altman’s own “hiccups” warning sits at the intersection of these narratives: a candid admission that the same explosive growth OpenAI is courting in the workplace could strain the systems meant to keep GPT‑5.6 and ChatGPT Work reliably available.