OpenAI News & Discussions

[wjfox]

ChatGPT uninstalls surged by 295% after DoD deal

https://techcrunch.com/2026/03/02/chatg ... -dod-deal/

I'm still using it for now. But growing increasingly frustrated and disappointed with Altman.

[wjfox]

Introducing GPT‑5.4

March 5, 2026

Today, we’re releasing GPT‑5.4 in ChatGPT (as GPT‑5.4 Thinking), the API, and Codex. It’s our most capable and efficient frontier model for professional work. We’re also releasing GPT‑5.4 Pro in ChatGPT and the API, for people who want maximum performance on complex tasks.

GPT‑5.4 brings together the best of our recent advances in reasoning, coding, and agentic workflows into a single frontier model. It incorporates the industry-leading coding capabilities of GPT‑5.3‑Codex⁠ while improving how the model works across tools, software environments, and professional tasks involving spreadsheets, presentations, and documents. The result is a model that gets complex real work done accurately, effectively, and efficiently—delivering what you asked for with less back and forth.

In ChatGPT, GPT‑5.4 Thinking can now provide an upfront plan of its thinking, so you can adjust course mid-response while it’s working, and arrive at a final output that’s more closely aligned with what you need without additional turns. GPT‑5.4 Thinking also improves deep web research, particularly for highly specific queries, while better maintaining context for questions that require longer thinking. Together, these improvements mean higher-quality answers that arrive faster and stay relevant to the task at hand.

In Codex and the API, GPT‑5.4 is the first general-purpose model we’ve released with native, state-of-the-art computer-use capabilities, enabling agents to operate computers and carry out complex workflows across applications. It supports up to 1M tokens of context, allowing agents to plan, execute, and verify tasks across long horizons. GPT‑5.4 also improves how models work across large ecosystems of tools and connectors with tool search, helping agents find and use the right tools more efficiently without sacrificing intelligence. Finally, GPT‑5.4 is our most token efficient reasoning model yet, using significantly fewer tokens to solve problems when compared to GPT‑5.2—translating to reduced token usage and faster speeds.

Together with advances in general reasoning, coding, and professional knowledge work, GPT‑5.4 enables more reliable agents, faster developer workflows, and higher-quality outputs across ChatGPT, the API, and Codex.

https://openai.com/index/introducing-gpt-5-4/

[weatheriscool]

[weatheriscool]

[weatheriscool]

[firestar464]

OpenAI is rethinking ChatGPT pricing — and 'unlimited' plans may not last, its boss says

https://www.msn.com/en-us/money/other/o ... r-AA1YPKas

[weatheriscool]

[weatheriscool]

[firestar464]

OpenAI is scrapping the Sora app to chase bigger AI goals

https://www.businessinsider.com/openai- ... ons-2026-3

[weatheriscool]

[firestar464]

ChatGPT's taste for literary nonsense sparks alarm

https://techxplore.com/news/2026-03-cha ... alarm.html

They should def test Claude, Gemini, and Grok on this too

[firestar464]

FT- OpenAI puts erotic chatbot plans on hold ‘indefinitely’

https://ghostarchive.org/archive/2jSeF

[weatheriscool]

[firestar464]

[firestar464]

[wjfox]

[firestar464]

[firestar464]

[Yuli Ban]

I keep asking myself sometimes "Why was such a tier not available before now?"
If anything, I'm glad. Back in February or March of last year, I was constantly trying to convince myself to buy a month of ChatGPT Pro, and figured that I'd get around to it but kept putting it off for different reasons. And thank f*ck I did! They depreciated the ONE ChatGPT model I ever unironically liked that wasn't 3.5 in 2022: o1. That original full model was genuinely pretty good; I'd be daring to say that the utility I got out of o1 has barely been exceeded if at all by current SOTA releases. Then they replaced it with that o3, which I called the "margins notes writer" model because whenever I got it to generate shitty creative writing slop to listen to, without fail it'd have characters write notes in margins or create bookmarks and whatnot while making some god awful "profound" quote. That's probably when I formally gave up on even thinking of AI slop for creative writing in any capacity and nothing I've seen since tells me I made the wrong choice. The benefits of this being I've fully embraced a luddite take on creative writing and have only become a stronger writer for it. 4.5 also released around that time, and I remember being SO hyped up for it and GPT-5, for years. Ever since GPT-4 dropped, I convinced myself that the "GPT-4.5 class" was going to be damn near proto-AGI, said that Claude 3 Opus was "GPT-4.25 class" and thought o1 was also in that same class but at its top end and that GPT-4.5 would be like o1 but multimodal with crazy long memory and whatnot, like it'd be able to one-shot generate whole novellas. o1 already came close to that, I remember one output analytical take was something like 11,000 words, and it wasn't even totally terrible (despite being ridiculously hallucinatory)
Then it actually released, it wasn't any better than 4o, and was so expensive that on the ChatGPT UI it had HALF the output window of 4o.

Again! Thank f*ck I put off buying ChatGPT Pro. I wouldn't have done so if there was a $100 option; that's about as much as I would be willing to spend for a premium LLM (and I even said that way back in 2023, remember, when OpenAI first made noises about a top-end option above Plus, that I'd not mind a $50 a month option and would even be willing to go as high as $80 or $100 a month, but anything above that is not going to cut it). Jumping from $20 to $200 was just too steep for virtually no real benefit. My only regret was that I never got to try o1-Pro but since they depreciated it, no skin off my nose.

Sad thing is, even if blowing $200 was nothing to me, as inconsequential as buying a 25¢ gumball, I still wouldn't be arsed to spend it on any current LLM now that their limitations are well known to me. Not even just to have no real limits. You'd need to have to give me a model that's everything I thought GPT-4.5 would be to convince me to try. I'm just so burned out on LLMs.

[firestar464]

"gimme novel" is the wrong approach (that was never gonna end well); it's better to use it for feedback.