> After that, a top goal for us is to unify o-series models and GPT-series models.
In summary:
* GPT-4.5 will be the last non-chain of thought model to be released.
* GPT-5 is (GPT-4.5 + o3) with multi-modal and reasoning.
* Plus subscribers will be able to run GPT-5 at a higher level of intelligence.
* Pro subscribers will be able to run GPT-5 at an even higher level of intelligence. These models will incorporate voice, canvas, search, deep research, and more.
If you are wondering when this will be released to (dis)replace a lot of economically viable jobs? This year. [0].
The big question should not be about when this will be released. It should be why is he announcing this *now* and what is Sam still not telling you.
You better start preparing for 2030 if I were you.
"weeks/months" I think referred to GPT-4.5 in weeks, GPT-5 in months
"It should be why is he announcing this now and what is Sam still not telling you."
- I don't think there is anything mysterious here? People constantly spam him on Twitter asking for roadmap/wen next release.
Yeah I also don’t get it why this is considered suspicious. Can someone please clarify?
It is also not that different compared to previous announce and release cycles other than gpt5 being their flagship model e.g. sora and o3 mini have been demoed weeks before the release
It’s hard not to frame these announcements in the context of broader competition like Deepseek or Musks bid to buy OpenAI.
The rate of progress and the rate of announcement and speculation are not the same I think and we should expect all kinds of marketing tactics as part of the discourse.
Does it really matter when they're gating their most powerful models with heavy limits on Plus users? You can barely use o3-mini-high for a couple of hours without getting limited for literal days.
So if you subscribed to "pro" because Sam said o3 and o3 pro would be available to ChatGPT pro subscribers by the end of February you are out of luck...
Great to see OpenAI acknowledge how difficult their stuff has got to use at this point:
> We want AI to “just work” for you; we realize how complicated our model and product offerings have gotten.
> We hate the model picker as much as you do and want to return to magic unified intelligence.
Using their stuff effectively right now requires you to pick between GPT-4o, GPT-4o with scheduled tasks, o1, o3-mini and o3-mini-high... and then understand which tools (Deep Research, DALL-E, Code Interpreter, Web search, Canvas) are available for which models and what each of those tools will let you do.
Well, different humans have different capabilities and we have to choose which one we want to talk to depending on what we are trying to accomplish. Therefore these tools are as good as or better than humans in every possible way. Stop being a hater. ;)
I think this means that the cost of cutting-edge models such as o3 pro has become too high, and that ordinary users can no longer use them freely.
I fear that, even though it is clear that smarter models are optimal, OpenAI will use cheaper models for its own convenience.
16 comments
[ 2.6 ms ] story [ 40.1 ms ] threadIn summary:
* GPT-4.5 will be the last non-chain of thought model to be released.
* GPT-5 is (GPT-4.5 + o3) with multi-modal and reasoning.
* Plus subscribers will be able to run GPT-5 at a higher level of intelligence.
* Pro subscribers will be able to run GPT-5 at an even higher level of intelligence. These models will incorporate voice, canvas, search, deep research, and more.
If you are wondering when this will be released to (dis)replace a lot of economically viable jobs? This year. [0].
The big question should not be about when this will be released. It should be why is he announcing this *now* and what is Sam still not telling you.
You better start preparing for 2030 if I were you.
[0] https://x.com/sama/status/1889757267425370415
"It should be why is he announcing this now and what is Sam still not telling you." - I don't think there is anything mysterious here? People constantly spam him on Twitter asking for roadmap/wen next release.
It is also not that different compared to previous announce and release cycles other than gpt5 being their flagship model e.g. sora and o3 mini have been demoed weeks before the release
The rate of progress and the rate of announcement and speculation are not the same I think and we should expect all kinds of marketing tactics as part of the discourse.
Also, I wonder if any of these decisions around GPT-5 are intended to make it harder for competitors to distill their models?
I am *really* hoping this is stronger than sonnet.
The future is bright /s
> We want AI to “just work” for you; we realize how complicated our model and product offerings have gotten.
> We hate the model picker as much as you do and want to return to magic unified intelligence.
Using their stuff effectively right now requires you to pick between GPT-4o, GPT-4o with scheduled tasks, o1, o3-mini and o3-mini-high... and then understand which tools (Deep Research, DALL-E, Code Interpreter, Web search, Canvas) are available for which models and what each of those tools will let you do.
And o1 will pretend to run Code Interpreter even though it doesn't have access to that tool yet! https://chatgpt.com/share/677420e4-8854-8006-8940-9bc30b7088...