GPT 5 Launches Today

Yesterday Sam Altman made a tweet that can only be understood as the launch date and time for GPT 5. GPT version 5 again is the most anticipated LLM in the world, and is believed to be "AGI level". If correct, this implies we now have a date and time for the singularity, and the time is 10AM today the 7th of August 2025, 8PM Cyprus time, and 6PM UK time.

GPT 5

GPT 5 is a "protected name". Since the release of GPT 4, OpenAI have released dozens of new models, but none of them qualified to become "number 5". This is because their 5th release have so huge expectations that even the differences between GPT4 and their latest model called GPT4.1 are expected to be "minor upgrades" compared to the differences between 4.1 and 5. So what can we expect from GPT 5?

I'm not going to go into speculations, but you can expect it to be the single largest and most important release from OpenAI since ChatGPT went viral. Maybe they'll solve hallucinations, or at least have significantly reduced the problem? Maybe a 10 million token window? Maybe the ability to run 10x "IQ" with 10% of the resources? Maybe integrated tools making existing integrations look like a "child's play"?

The only thing we know for sure, is that today's GPT 5 launch from OpenAI will be massive!

Since this is the most carefully planned and orchestrated release they've probably ever done, we can assume the improvements will be massive.

AGI?

AGI is a politically loaded word. According to OpenAI's charter they cannot protect "AGI level intelligence" using their existing means. Their own foundation dictates that it should be available as open source, indiscriminately to the world at large, making the announcement of "AGI-level LLMs" from OpenAI a difficult medicine to swallow for themselves.

GPT 5 might therefore very well be "AGI-level intelligence" without that ever being announced

At AINIRO for the record, we believe we've collectively had access to AGI for roughly 1.5 years now to put things into perspective. It all depends upon your definition of "AGI", and "AGI" is not a definition the sector collectively agrees upon.

Wrapping up

This article is just speculations, but with the single largest and most anticipated LLM release happening in 11 hours from now, I assume most others are also speculating. For years there's been speculations about this release, and the common concensus is that it'll be OpenAI's "enter the AGI era model". From AGI of course, the next step becomes ASI - Which in theory should be easier, since an AGI model should be able to recursively self improve, beyond what humans can make it do - Implying we'll get a recursive self feeding improvement loop, that just automagically spits out new and improved LLMs, given electricity as its only input.

However, what do you think? Is this just another LLM from OpenAI? will it become worse than their existing models? This has happened previously. Will it be their best model by far? Will it be AGI, or maybe even ASI from the get go? Tune in, 10PM Pacific Time today, and let's see ...