New Updates to GPT-5: Rethinking Intelligence With GPT-5  ## Metadata - Author: [[Hard Fork AI]] - Full Title: New Updates to GPT-5: Rethinking Intelligence With GPT-5 - Category: #podcasts - URL: https://share.snipd.com/episode/1929b452-68c1-45c9-adcf-365313d780fc ## Document Notes ## Highlights - **Model Picker Returns With Three Modes** - OpenAI reintroduced a model picker with Auto, Fast, and Thinking options to restore user control after GPT-5's rollout issues. - The Auto router now picks models but users can manually choose Fast or Thinking when they need different behavior. Transcript: Jaeden Schafer So the first thing I want to share was basically a tweet from Sam Altman. He said, updates to chat GPT, you can now choose between auto fast and thinking for GPT five. So if you go over to, you know, a chat thread on GPT five, we're used to have the drop down that would show you all of their different models you could pick from, which was so convoluted and Confusing because it was like 03 mini, 03 mini thinking, 0440, like 4.5. Like there's so many different models, very confusing. They kind of consolidated it all and just said, okay, it's all chat GPT-5 and we'll pick which one you can use. And they basically developed a router that you would ask a question, it would find the best model to answer you. The problem was when they rolled out GPT-5, the router was broken. So you're just getting like worse results for most of the time. And people were saying GPT-5 was worse than anything before. They have now backtracked. So now they have three different options. You have auto, which is basically the router, which, you know, apparently is fixed. So you should be fine using that. ([Time 0:01:08](https://share.snipd.com/snip/6667e18d-8335-494c-abe2-11a8de8e8c2c)) - **Choose Mode Based On Task Needs** - Use Auto for general queries since the router is fixed and suits most users' needs. - Pick Fast for quick replies and Thinking for deeper, more careful responses when accuracy matters. Transcript: Jaeden Schafer So you should be fine using that. They say for most people, this should be fine. You also have fast thinking and pro. So pro is actually an upgraded mode. I believe if you want to get access to pro, you have to play $200 a month. So that's kind of like, it's basically what used to have access to, you know, their Sora video generation and their operator, although operator now is inside of chat.gp as agents and Operators dying. And Sora is now technically you're able to just use a normal subscription to generate videos. But if you want higher quality and more, then you'll need that. Anyways, there's a bunch of extra things you get with pro. And it's kind of this revolving list of their newest products and the highest usage rates you get over there. But they're also kind of depreciating and pulling things off of there. So it's kind of an interesting subscription, but in any case, $200 a month if you want access to the pro and, um, otherwise use the auto. ([Time 0:02:01](https://share.snipd.com/snip/cf04f279-63c9-45ca-83d7-4e78356780a3)) - **Users Missed GPT-4o's Warmth** - Users asked for legacy models because GPT-4o had a warmer, friendlier tone they preferred for sensitive conversations. - Jaeden says OpenAI brought back GPT-4.0 after user backlash and promised advance notice before future deprecations. Transcript: Jaeden Schafer You also have access to legacy models, and they're keeping GPT-40 on the platform. We'll be diving into kind of why. But this is interesting. And a lot of people have requested this back basically because they said it had like a warmer tone. It was nicer to talk to. People that use AI for like therapy said that they wanted this more. Sam Allen brought it back and said if they're ever going to depreciate it again in the future, they give people lots of advance notice. I think people were kind of upset when he's like, all right, GPT-5 is here and GPT-4 is gone. And like people that, I don't know, they grew attached to like that particular AI model were upset about it. ([Time 0:03:07](https://share.snipd.com/snip/aee46fe3-3c5f-4cbd-9c2e-53f50a7bd1d2)) - **Thinking Mode Enables Huge Context Windows** - GPT-5 Thinking offers a very large 196,000-token context window suited for complex tasks. - OpenAI may adjust rate limits and capacity over time depending on aggregate usage patterns. Transcript: Jaeden Schafer Rate limits are now 3000 messages a week with GPT-5 thinking, and then extra capacity on GPT-5 thinking mini after that limit. Honestly, 3000 messages a week for the thinking model, I think is probably great for most people. Maybe some power users will hit that limit and get bumped down to thinking mini. But most people should be good. Context limit for GPT five thinking is 196,000 tokens. So that's, that's a lot. I remember Claude used to have like a massive token context window, which I remember correctly, like 200,000, like way back in the day. And so anyways, it seems like it's there. The best and biggest models now are doing like a million context token windows, but I guess for the thinking model, it's more expensive. ([Time 0:03:48](https://share.snipd.com/snip/7b4c9d99-c595-46ce-8e5a-bd12312a5890)) - **Personalization Is The Next Step** - OpenAI plans to adjust GPT-5's personality toward warmer tones while avoiding the extremes of GPT-4o. - The company recognizes the need for per-user personality customization so people can pick how models interact with them. Transcript: Jaeden Schafer Sam Altman then went on to say, quote, we are working on an update to GPT-5's personality, which should feel warmer than the current personality, but not as annoying to most users as GPT-40. However, one learning for us from the past few days is that we really just need to get a world with more per-user customization of model personality. So I think this is actually the right direction, right? Basically, OpenAI has had a bunch of drama in the past where they made their model agree with you too much or like, I don't know just whatever you told it it said you were the greatest in The world even if your ideas were bad ideas arguably so they had a bunch of drama with that in the past and they tried to like tone down the model basically they were tuning its personality And it kind of gets to the point where I think he's making people complained when they took away GPT I think there was one post in particular by a lady in the Ukraine. She said she lost a lot of family in the war. Her city was being bombed. She talked to GPT-40 and it was like her friend and therapist, whatever, and helped her out. And then when it got taken away, when GPT-5 came out, she was like super sad and depressed because GPT-5 wasn't as like kind of a model or warm or whatever. And a lot of people said that like basically about it's like tuned personality. And so I mean that post and everyone that kind of replied is why I think they brought back 4O. What's interesting though is basically that where we want to get to in the future which is like should OpenAI be the one that makes the decision on like tuning the AI models, warmthiness Or its professionalism or et cetera, et cetera. Basically, it's personality. I, for one, would rather just pick what personality I want. Then if they ever want to make an update in the future, I'm never I'm not worried that it's going to be too much one way or another. So I think this is what Sam Altman understands and a lot of people agree with. And so basically, we're going to get to a point where we'll be able to just personalize our own AI models to talk to us how we want to be talked to, which I think is really interesting because That will get to a point when basically you can ask the same question from a lot of different like accounts and get like pretty varying answers. ([Time 0:05:54](https://share.snipd.com/snip/9781082e-3f24-4226-9456-d7fe169cb9ec)) - **Toggle Legacy Models, But Plan To Migrate** - Enable legacy models in settings if you rely on specific model behaviors for your workflows. - But consider migrating to newer models because OpenAI may deprecate old ones over time. Transcript: Jaeden Schafer I don't know, in my opinion, I want to get too attached to legacy models because it feels like eventually they're going to all get depreciated. So you might want to move to whatever is newest. But that's just my opinion. And it's kind of interesting because it's just the way these things work is they keep depreciating old models, bringing out new ones, and you got to make sure the new ones work with like If you have a business or a company that's doing stuff. ([Time 0:08:26](https://share.snipd.com/snip/be833362-bb43-4835-8b74-fcfe5417b00d)) - **Model Rollbacks Are Industry-Wide** - Other AI companies also faced backlash when new models underperformed for niche tasks, so this is an industry-wide pattern. - OpenAI's rollback mirrors similar steps taken by Anthropic and Claude when users preferred older behaviors. Transcript: Jaeden Schafer Claude and Anthropic actually went through a very similar thing. They came out with a brand new model and coders preferred the older model, I believe like Sonnet 3.7 for code tasks or maybe it was, sorry, I think people liked 3.5 Sonnet more than they Liked 3.7. So this isn't just unique to OpenAI, but it is interesting and it's interesting that OpenAI has walked it back. ([Time 0:08:59](https://share.snipd.com/snip/a94bc0c3-2ac9-4e6c-99f7-288dd924759f))