I think when GPT-5 is released it will probably be so ahead that they will evaluate it on new benchmarks due to the new capabilities we didn't expect.
Don't underestimate OpenAI at all.
I think gpt-2 is likely a much smaller model for on device purposes.
We've seen models get smaller and more efficient (Phi-3, llama 3 etc)
OpenAI are secretly developing a new product which will require an on device A.I.
I think this makes sense?
Sam altman and greg brockman have clearly stated that they will "literately" deploy GPT -5 and future models.
This is literally them saying.
We aren't realising everything we have all at once because you literally aren't ready for what's coming.
Which actually means they are
Prediction: By end of 2024 we will see
• 7-10 GPT-4 level models
• No massive advance (no GPT-5, or disappointing GPT-5)
• Price wars
• Very little moat for anyone
• No robust solution to hallucinations
• Modest lasting corporate adoption
• Modest profits, split 7-10 ways
We're releasing a new iteration of SWE-bench, in collaboration with the original authors, to more reliably evaluate AI models on their ability to solve real-world software issues.
I think OpenAI are actually thinking how can we NOT shock people with the updates.
Hence taking so long to post progress and when they do its incremental updates.
People acting as if A.I images don't influence how you see someone even if you KNOW the image is A.I generated.
The image on the right doesn't exist yet people think he's "cooler".
I think that GPT4o voice mode isn't being rolled out yet because once it does, it will reach max capacity very quickly.
That kind of technology will really be used a lot more than you think.
Another chatgpt moment but for voice.
Introducing Maisa KPU: The next leap in AI reasoning capabilities.
The Knowledge Processing Unit is a Reasoning System for LLMs that leverages all their reasoning power and overcomes their intrinsic limitations.
I said this in a video somewhere.
Not sure if my video has released yet since i made a bunch.
But with increased capabilities they will NOT be giving access those advanced models to everyone.
The risk is too high.
You will 100% need a certain id/license to use certain AI
GPT-5 Might be make or break.
Either we will see if these models get truly better at reasoning and coherence or people like
@GaryMarcus
will be proven right.
We will see if its all hype or there's truly exponential growth coming.
I think that GPT4o voice mode isn't being rolled out yet because once it does, it will reach max capacity very quickly.
That kind of technology will really be used a lot more than you think.
Another chatgpt moment but for voice.
Google just dropped an elite AI mathematician.
It's a neuro-symbolic system that formalizes problems into Lean, a formal language, with a fine-tuned Gemini and uses AlphaZero-style search to solve them. Solves 4/6 on IMO 2024, a Silver medal.
The future of Math be AI-assisted.
I see so much gripe about whether AI can do XYZ.
It might not be able to do certain things today.
But then suddenly, just like SORA.
A new threshold will be reached.
Many people are in denial.
Everyone is stating that this makes no sense.
Well it actually does.
GPT5,6,7 are all going to be massive improvements in coding.
And because Devin is an advanced GPT-4 wrapper...
This product will get MASSIVELY better over time.
They probably already have tons of
One of the things that confuses me so much about A.I
The smartest people in field are not on the same page.
Makes it so much harder to predict the future.
I wonder what was goanna be released but Open AI Decided against it?
They recently trademarked "SORA" one day before release on Feb 14th.
They trademarked GPT-V the same time leakers were showing a page called 'v-test'
My best bet is that the competition isn't anywhere near
So far I think these 2 members that are taking right now are both bots.
THE BEST intelligence i have seeen so far!
If this is a human I will be very surprised, and if this is a robot well done I will take this as gpt 5 anytime!
Can you tell?
The gpt-2 chatbots are back and all indications are these are indeed the latest GPT-5 versions
They seem to be superior to Opus and GPT-4, but still quite underwhelming compared to the insane hype about GPT-5
It will be hilarious if Gemini 2.0 and Llama-3 400b beat it!
I also don't think that they are going to cold drop GPT-5.
The last time they had an update, we knew the date in advance.
They could build so much hype around GPT-5 , not sure why they would waste it.
Nobody thought they stole her voice.
They just don't like that she said no twice and they used someone extremely similar.
And then referenced the movie multiple times?
2023: $30,000 to get a simple iPhone app created, $300 for a plumbing job.
i wonder what those relative prices will look like in 2028!
the likely coming divergence between changes to cognitive work and changes to physical work could be quite dramatic.
If OpenAI were previously an Open company
And now they are closed for profit...
Supposedly working to "distribute agi"
How do we knowfor 100% they wont change once again after AGI and ASI?
Genuine Question.
GPT-4o hot take:
• The speech synthesis is terrific, reminds me of Google Duplex (which never took off).
but
• If OpenAI had GPT-5, they have would shown it.
• They don’t have GPT-5 after 14 months of trying.
• The most important figure in the blogpost is attached below. And
> yeah it cant make images properly
> yeah they dont look photorealistic
> yeah they are photorealistic but look at the hands
> yeah the hands are perfect but it cant do video
> yeah video looks good but look at the morphing
Its not wise to underestimate this trend.
A GPT-4 level chatbot, available to use completely free, running at over 800 tokens per second on Groq.
I'm genuinely mindblown by LlaMA 3.
Try it with the link in the next tweet.
They said they aren't releasing it.
"Given such context, we have no plans to release an online demo, API, product, additional implementation details, or any related offerings until we are certain that the technology will be used responsibly and in accordance with proper
@Devgymsetvfx
i want to believe it think perhaps they even have an agi like system, im sure of it. But i just dont think Open AI even announces let alone releases that tech in 2024, without launching GPT-5 First, or 4.5.
Im sure they are way ahead of what everyone else has.
SORA backlash
bro dalle 3 getting wild wtf. why is it so good? these dont even look ai generated they look like theyre from a movie or something. dalle 3 always getting iteratively better, openai just never says anything about it.
Again, im not an AI leaker at all. Its publicly available information that could be 100% wrong.
But since OpenAI have stated that they will deploy models iteratively...
And the fact that the sora trademark was very accurate
@SmokeAwayyy
I'm leaning towards it being more
As more and more AI Content is online
People will believe less of what they see
Ai content wont be the problem
Getting people to believe real things will be...
Nobody will believe anything anymore.
Yikes
Greg brockman leaving..
Now an updated graphic showing a marginal increase as opposed to a huge one...
Claude 3.5 sonnet infront....
Not looking good
But then again this is speculation
Will be a lot of tweets regarding research papers...
There's actually a lot of information that just gets lost.
So will be posting more on here starting today.
Yes because people who cover AI news are not going to talk about the fact that the CEO of OpenAI publicly responded to an account tweeting about their exclusively secret model.
I've trained a custom model on the OpenAI Sora alien walking in NYC.
The results are just WILD 🤪
See for yourself 👇👇👇 (and learn how to unlock the same magic on )
Q*
> can solve grade school math problems
> so what, what’s the big deal?
> why did the researchers send the letter ?
> why did Ilya freak out
Because OpenAI already has a stack where they can predict intelligence based on compute and data. These disclosed this with the GPT4
Admire entrepreneurs like
@adcock_brett
who don't just take money from first exists and chill on a beach somewhere. Hes pushing again with
@Figure_robot
Impressive stuff
Kids are in bed, heading back to the office again
Oh btw, when I started Archer - biggest pushback I received early on “you’ll need a billion dollars or it will die”
5 aircraft versions later and now $1.3B of capital…
Let’s push
As crazy as Sora is , it doesn't change that much in the grand scheme of things.
Agents + AGI will be the true eye opening moment for everyone.
We literally have state of the art LLMs and yet there are people who can write better than some LLMs and still have careers in
I stand corrected: GPT-4o does NOT natively process video stream. The blog says it only takes image, text, and audio. That's sad, but the principle I said still holds: the right way to make a video-native model efficient is to co-develop the streaming codec on edge device.