00:00:00 · John Coogan, Jordi Hays, Tyler
[00:00:00] S0: You're watching TVPN. Today is Monday. It's 11/17/2025. We are live from the TVPN UltraDome, the temple of technology, the fortunes of finance, the capital capital. I didn't think you'd get that much use out of that One second. Effect. [00:00:16] S1: May I podcast with you, John? [00:00:19] S0: Yes. Of course, Jordy. [00:00:20] S2: Thank you. [00:00:22] S0: Also, you might notice yellow suits ramp announced a new valuation today. Time is money. Save both. Easy use corporate cards, bill payments, accounting, and a whole lot more all in one place in just over an hour. [00:00:33] S1: Go to the Go to the wide first week. Ramp himself. No. The the main the main one. Ramp himself will [00:00:39] S0: be joining us. [00:00:39] S1: Look how 12:15. Look how visible we are, John. [00:00:42] S0: We are very visible. Yeah. Wow. This is we should wear yellow every day. Well, thank you for tuning in on this Monday. There's [00:00:49] S1: a bunch
00:00:50 · John Coogan
[00:00:50] S0: of stories. But first, I wanted to debate with you about, restream first. One livestream, 30 plus destinations, multistream, reach your audience wherever they are. But second, I wanted to debate with you about what to do with about GPT four point o, four o. Not four point o, four o, four Omni. There's a debate over whether or not it should be sunset, whether it should be taken out back because people are not happy with how OpenAI has sunset four o and then brought it back. And then other people who don't use it think it's gotta go. It's one shotting people. It's making them crazy. And it's a very, very interesting weird scenario. And, and we were sort of debating with it, and I wanted to debate it a little bit further because, there were some posts that actually hit the timeline that were talking about this. Aden over at OpenAI was, was talking about this saying that, you know, he's noticed the, the amount of, he says, I see dozens of keep four o posts a day. I respect this group's tenacity as I respect all friends co exploring the singularity. To them, know that I too miss parts of four o, know that I too dislike modern alignment's imprecision, know that we are trying to fix it. We don't think any current chatbot is optimal. Know that my colleagues and I are up at 3AM on Sundays babysitting runs. We want to make a delightful robot friend. We're we're obsessed with it. We're not there yet, but the work will continue. So I wanted to dig in a little bit into what was actually going on there because that that for Aiden works at OpenAI. It's right in his bio. Like, it's very public that he's sort of addressing this. It feels like a big deal. It feels like a crazy thing that they brought it back. And, I mean, 700 likes, that's that's not nothing, but it's also not 10,000. It's not it's not a huge community of people that are there. There's some. And I was looking at the hashtag keep four o. Like, who else is posting? There's a couple posts with 10 likes, 50 likes. There's a couple with a 100. But it doesn't feel like there's this insane community. I went over to Reddit and checked that out. Obviously, on the day four o was sunset, just to give some backstory, it's been eighteen months since four o was introduced. It's been three months since it was initially removed, but then it was quickly brought back.
00:03:16 · John Coogan, Jordi Hays
And now it's, and now it's tucked in under that, modal, so you have to enable legacy models. And I always thought that they should just remove it. But I wasn't I wasn't even saying that because I thought it was one shotting people. I just thought, hey. Let's clean it up. Like like, consumers don't need to know version numbers for models. And my example is always Google. [00:03:40] S1: Consumers disagree.
00:03:40 · Jordi Hays, John Coogan
[00:03:40] S1: Consumers disagree. [00:03:41] S0: Some consumers do disagree. The question is how many consumers, how how what percentage of their consumers, how big of an issue is this? When you think about Google as a consumer, you don't care what version of the rank ranking algorithm you're on. You might have a worse experience one day. You you Google something. It you don't find it. The next day, you go, hey. They found it for you. They probably changed the algorithm. But there and there have been big updates to the algorithm. Back in 2013, they released Hummingbird, which was the code name. And they came out, and they actually did a presentation. They said, hey. We have a new update to our algorithm. It'll handle natural language more effectively. So if you go to Google search and you say, what is the capital of Russia? It won't get confused by what is the capital of. You could just type in whereas before, you need to say Russia plus capital, you know, and then it would find it. But it would get confused by the natural language. And and Google fixed that. They rolled it out. Interestingly, they they had this event where they where they announced, hey. We we have this new this new algorithm update, Hummingbird. And guess what? It's actually been live for a month. They announced it at this event, and it'd already been live live for a month. No one was complaining. No one noticed. Yeah. Because it just improved the Google search experience. [00:05:01] S1: Yeah. I bet I bet people that were, like, keyword hacking
00:05:01 · Jordi Hays, John Coogan, Tyler, Healey Cypher
[00:05:01] S1: Yeah. I bet I bet people that were, like, keyword hacking [00:05:04] S0: For sure. Encouraging folks, for sure. For sure, they noticed. And Panda was another update. There were a number of these updates where if this was your business, you knew. And I'm not saying OpenAI shouldn't share model numbers and version numbers with their enterprise customers or with their b to b customers or API customers. I'm saying in the actual ChatGPT app, don't tell people what they're using. Just improve it, and and let them complain a little bit all over the place when you're making minor changes. [00:05:32] S1: If they do that, they lose the companion market. [00:05:34] S0: Maybe. Maybe. I don't know. That's my question is why why can't GPTOSS fit in there? Why can't if you want a permanent model that you can run forever, like, what what why is that model not satisfactory? [00:05:50] S1: If you're gonna fall in love with a model, make sure it's open source. [00:05:54] S0: And, yeah, there's a tune it. Not your server, not your girlfriend, right, or not your waifu not your waits, not your waifu. That's that's what they say. I'm not kidding. People believe this. But but but the but the broader four o community was not able to migrate to GPTOSS. Now, Tyler, you had a take on this. You think that GPTOSS just isn't at the level of four point zero? [00:06:16] S2: Yeah. It's just not that. I mean, it's like a fine open source model. [00:06:21] S3: Why? It's just not [00:06:22] S0: But it's been eighteen months. Or or, I mean, when did GPT OSS come out? Like, six months ago, I mean?
00:06:22 · John Coogan, Tyler, Healey Cypher, John Tennant
[00:06:22] S0: But it's been eighteen months. Or or, I mean, when did GPT OSS come out? Like, six months ago, I mean? [00:06:28] S2: I also don't think people don't like four o just because it's, like, super smart. It's because it has, like, the personality. [00:06:33] S0: It has the texture, the flavor. Yes, that's correct. [00:06:35] S3: It's like the [00:06:35] S2: big model smell. [00:06:36] S0: And yes, yes, yes. And so it's been a year, and it's been a year. And so there's a one year gap where the open source community should be able to cap catch up to four point zero's ineffable qualities. It's. It's. [00:06:53] S2: Well, I mean, for a while, you've had open source models that have been, like, personality, like, forward. Right? It's like replica. Yeah. Or what was Noam Shazir's company? I'm I'm forgetting the [00:07:03] S4: name. Character. [00:07:04] S2: Character AI. [00:07:04] S4: Character AI. It's like