GPT Monster Morphing Muddle

Fri, 10 Jul 2026 12:33:34 +1000

As OpenAI try to move forward their models and apps they’re creating quite the muddle. My impression is that OpenAI are trying to do too much, too fast and in the process burning their users.

This morning (or overnight my time) they’ve rolled out

New Family of models with GPT-5.6 in three variants: Sol (considered its workhorse), Terra (a more intermediate option), and Luna (its budget-friendly option)
New concept “Work” - Use Work when you want ChatGPT to complete a task with a clear outcome, such as a brief, deck, analysis, recurring update, workflow, or file you can review and use.
New apps for Mac, iOS and Windows - chatgpt.com/features/…
New web experience to match the “Chat”, “Work”, “Code” switch
New “Sites” in Codex - Build and share hosted sites in ChatGPT. Sites lets ChatGPT create, host, refine, and share websites, web apps, and games. Use Sites when you want to turn a prompt or compatible existing project into a hosted experience without setting up a separate deployment workflow. Demo of Sites via YouTube
Introducing GPT‑Live: “A new generation of voice models for natural human-AI interaction, now powering ChatGPT Voice” - a seriously impressive new voice interaction model which dropped a day earlier.

The rollout is very uneven. At present I can see “Work” in web and iOS but not on desktop. I can’t see “Codex” anywhere.

OpenAI are moving to the same position as Anthropic (Claude) with one app and the three modes; Chat, Work and Code.

Use Chat when you want an answer, explanation, brainstorm, or short draft. Use Work when you want ChatGPT to complete a task with a clear outcome, such as a brief, deck, analysis, recurring update, workflow, or file you can review and use. - learn.chatgpt.com/docs/use-…

Work is a good step forward but replicates what could previously be done with Codex app. It was a solid, native app that beat Claude Cowork for speed and slickness. For example, it embedded a fully functional Web preview in the app.

It sounds like the quality of this native app has been replaced by a lumbering Electron app - a full browser in a pretty wrapper. John Gruber pulls no punches when he says, “Today’s the Day OpenAI F*cked Up the ChatGPT Mac App”

OpenAI have said more than 5 million people use Codex each week, with more than 1 million using it outside software development.

Millions of users! Why would they be so cavalier in this rollout?

It’s a shame really as the performance of the new GPT-5.6 models is very good when considered on a value for money basis. I can access them today through web and iOS but my Mac is still stuck waiting for the new app to drop.

It’s just lucky I wasn’t racing fast to meet some deadline only to be derailed by this appalling lack of planning and change management.

Fuming - ChatGPT just dropped the ball

Fri, 10 Jul 2026 11:01:29 +1000

OpenAI just broke something that was working fine, in the name of progress. Software developer’s first rule, like doctors, should be “first do no harm.”

I’m a paying subscriber. I’ve been running Codex hard lately - finance modelling work, not so much straight coding - and my estimation of GPT’s capabilities had genuinely grown over recent weeks. Good performer. I was impressed. But not any more.

This morning I heard rumours of GPT 5.6 dropping, so I updated both the ChatGPT desktop and the Codex desktop app. Codex then refused to deliver a working application. After some fossicking, I found the release note:

On July 9, the Codex app merges into the new ChatGPT desktop app for macOS and Windows. Codex keeps its dedicated coding experience alongside Chat and Work, with inline editing in diffs, pull request review in the side panel, faster Computer Use powered by GPT-5.6, and multi-repository projects.

What a clumsy, abrupt change. They killed a familiar workflow with no warning, and the replacement isn’t even live yet.

Yes, they’ll argue it’s a staged global rollout. That argument doesn’t hold. It is simply not acceptable for a commercial application to go dark mid-transition and leave users waiting hours - or days - for restoration. “We’re rolling it out progressively” is not a service level.

And the timing argument is worse. The cutover happened “at night” - nighttime in the USA. It’s primetime here in Australia, and I cannot do the work I had planned this morning. That’s not a minor inconvenience; that’s a broken service contract.

If OpenAI expects business users to stake their workflows on their platform, this is appalling. Full stop, no hedging.

Fortunately I have Claude as a fallback today. And honestly, this whole episode is just accelerating my thinking about building more of my own AI stack - one that doesn’t leave me at the mercy of unreliable mega-corps doing surprise surgery on production tools.

It’s not just me who’s appalled and damaged by this cavalier treatment of paying customers.

Rambling Rows

GPT Monster Morphing Muddle

Fuming - ChatGPT just dropped the ball