In partnership with
OpenAI released GPT-5.6 on 9 July, a new three-model family, and paired it with ChatGPT Work, an agent built for long-running, multi-step tasks. In the same move it folded Codex into a redesigned desktop app that brings Chat, Work and Codex into one workspace on macOS and Windows.
The launch only came after a fortnight of federal gatekeeping. In late June the White House asked OpenAI to hold GPT-5.6 back to a small set of government-approved partners, vetted customer by customer, on the grounds that Sol's coding, biology and cybersecurity capabilities were "Mythos-like."
What can it do?
GPT-5.6 arrived as a three-model family across ChatGPT, Codex and the API
ChatGPT Work runs long tasks in finance, data analytics and engineering
A new desktop app unifies Chat, Work and Codex on macOS and Windows
Codex becomes a command centre with Skills, Automations and Worktrees
Start and monitor Codex jobs on remote machines from your phone
Partner message
Become An AI Expert In Just 5 Minutes
If you’re a decision maker at your company, you need to be on the bleeding edge of, well, everything. But before you go signing up for seminars, conferences, lunch ‘n learns, and all that jazz, just know there’s a far better (and simpler) way: Subscribing to The Deep View.
This daily newsletter condenses everything you need to know about the latest and greatest AI developments into a 5-minute read. Squeeze it into your morning coffee break and before you know it, you’ll be an expert too.
Subscribe right here. It’s totally free, wildly informative, and trusted by 600,000+ readers at Google, Meta, Microsoft, and beyond.
Why does it matter?
After an intervention, the Commerce Department's Center for AI Standards and Innovation eventually cleared it, but the precedent stands: Washington now decides when a frontier model reaches the public, and it is doing so without any framework saying it can.
Bundling a frontier model with an agent and a unified app pushes OpenAI from answering questions to finishing work, the clearest sign yet that agents, not chat, are the product.
An assistant that stays with a project for hours and acts across files and machines changes what teams buy software for, and raises fresh questions about oversight and security.
Our take
The consolidation matters more than the version bump. By merging Codex into the desktop app and placing work beside chat, OpenAI is building a single surface where a model does not just draft but runs. That is a play for the workday itself, competing less with search and more with the software companies rely on to get things done. TechRadar reported a backlash from users unhappy with the redesigned workflow, a reminder that consolidation forces old habits to change before the benefits land.
But agents that touch local files, browsers and remote hosts are useful precisely because they act with less supervision, which is also what makes them risky in regulated work. OpenAI is shipping scoped credentials and controls, yet the same week's Apple lawsuit shows how quickly hardware and IP disputes can shadow the company's ambitions. The product story and the legal story are now moving in parallel.
And another big thing… Apple sues OpenAI over hardware secrets
On 10 July Apple filed suit in the Northern District of California, accusing OpenAI and two former Apple employees, hardware chief Tang Tan and engineer Chang Liu, of misappropriating trade secrets tied to unreleased devices, parts and supplier relationships.
Apple alleges Liu used an authentication bug to download confidential files after leaving, and it is seeking an injunction and damages. OpenAI says it has no interest in anyone else's trade secrets. The case pits two of tech's largest players against each other as OpenAI pushes into hardware.
Go even more Agentic on YouTube
Last week’s video was all about Enshittification. Check it out.
