GPT-4, every mention
96 scenes, the whole family · ← back to GPT-4
tap a year for its mentions
every year anyone Emad Mostaque 16Alex Wang 13Douwe Kiela 12Arvind Narayanan 8Sam Altman 7Matt Clifford 6Ethan Mollick 5Richard Socher 4Max Junestrand 4Gustav Söderström 4
Verbatim, from the transcripts: the passages where GPT-4 comes up
Open Models vs Frontier Models: Who Actually Wins? | The $100K Token Budget Every Engineer Will Need · 20VC with Harry Stebbings
- ▶ 8:39 Clay Bavor That said, there will be an assembly line of, cool, uh, GPT-IV, which in, you know, March, April, May of twenty-twenty-three was good enough to do some set of things is now one 300th.
OpenAI vs Anthropic vs Open-Source | Token Maxing, AI Hangovers & The Coming ROI Reckoning · 20VC with Harry Stebbings
- ▶ 13:15 Matan Grinberg Like, just how before it was you know, GPT-II, then GPT-III, then GPT-III, then then GPT GPT-III then-III, then GPT then GPT then GPT GPT then GPT GPT GPT GPT GPT GPT-IV then,, then GPT then GPT then GPT-IV, then GPT, then GPT, then-
OpenAI & SpaceX S1 Drops | Layoffs at Cloudflare & ClickUp | OpenRouter & Polsia Raise Mega Rounds · 20VC with Harry Stebbings
- ▶ 22:24 unnamed speaker And as folks are a struggle to manage it themselves, that will, that may degrade pricing power for anthropic as folks just, you know, like, you know, I can, I'm just going to use a combination of GPT for it's older and some sonnet and some…
Replit CEO: Why the SaaS Apocalypse is Justified & Why Coding Models are Plateauing | Amjad Masad · 20VC with Harry Stebbings
- ▶ 26:38 Amjad Masad But like everyone's spending way more on opens than they were spending on GPT-IV, but they're getting way more intelligence
a16z, Anish Acharya: Is SaaS Dead? Do Margins Still Matter? Why We Are Not in an AI Bubble? · 20VC with Harry Stebbings
- ▶ 58:52 Anish Acharya So granted, closed is more expensive than open in many cases, but the cost of actually a token on GPT-IV has, you know, gone down a hundred X since the model was released.
Harvey CEO Winston Weinberg: How to Make Mega Deals | Lessons from Rabois, Halligan & Grady · 20VC with Harry Stebbings
- ▶ 22:12 Winston Weinberg Like, I, I feel like this is something that folks don't quite understand is like a lot of the consumer use cases, like four was good.
Benchmark's GP, Everett Randle on Why Mega Funds Will Not Produce Good Returns · 20VC with Harry Stebbings
- ▶ 29:50 Everett Randle Um, so I think when you think about like Jasper at first, the issue that they ran into was that, you know, people, when GPT four came out, they started being like, wow, like the outputs I'm getting from Jasper are kind of the same that I'm…
General Catalyst CEO, Hemant Taneja: Lessons Scaling GC to $40BN in AUM · 20VC with Harry Stebbings
- ▶ 31:53 Hemant Taneja Take customer support, for example, when new models come out, uh, when you, if you started building your company, when you were in the chat, in T three era versus four versus five, you just have more tools at your disposal disposal.
ElevenLabs CEO/Co-Founder, Mati Staniszewski:The Untold Story of Europe’s Fastest Growing AI Startup · 20VC with Harry Stebbings
- ▶ 10:26 Harry Stebbings And, you know, GPT-V was the embodiment of that.
20VC: 15 Term Sheets in 7 Days and Choosing Benchmark | Harvey vs Legora: Who Wins Legal and How to Play When You Have $600M Less Funding | Are AI Models Plateauing Today | Building a 9-9-6 Culture From Stockholm with Max Junestrand
- ▶ 15:45 Max Junestrand You had this company that had, as some others also did, pre-access to GPT-IV, which was an incredibly unfair advantage in some ways, but we started in a much smaller market.
- ▶ 37:10 Max Junestrand There were frankly a lot of firms who had piloted them back then who had an awful experience and who thought GPT-IV was better. 3 times in the scene
Sam Altman's Masterplan or a Gift to Anthropic? Palantir & Shopify Crush Earnings · 20VC with Harry Stebbings
- ▶ 6:31 Rory O'Driscoll They were focused on GPT for the following March.
Duolingo Co-Founder, Severin Hacker: How AI Impacts the Future of Work and Education · 20VC with Harry Stebbings
- ▶ 3:19 Severin Hacker So we were one of the launch partners of OpenAI when they first launched GPT-IV
20Product: How Scale AI and Harvey Build Product | Why PMs Are Wrong: They are not the CEOs of the Product | How to do Pre and Post Mortems Effectively and How to Nail PRDs | The Future of Product Management in a World of AI with Aatish Nayak
- ▶ 41:04 Aatish Nayak I think that whole paradigm was created for the, the, the Silicon Valley or like tech audience, because everyone loved playing around with like, oh, one versus GP four versus, you know, these models.
Kevin Scott, CTO @ Microsoft: An Evaluation of Deepseek and How We Underestimate the Chinese · 20VC with Harry Stebbings
- ▶ 35:57 Kevin Scott Like we, we have, uh, we have been running a thousand miles an hour building infrastructure over the past, uh, you know, two and a half years since GPT four.
Mike Krieger, Instagram CoFounder & Anthropic CPO: Where Will Value Be Created in an AI World?|E1265 · 20VC with Harry Stebbings
- ▶ 44:11 Mike Krieger I don't know if it still is, um, because I saw the potential and then, uh, even GPT four came out before it, they had multimodal and I was trying to do Swift with it.
Jonathan Ross: DeepSeek Special - How Should OpenAI and the US Government Respond | E1253 · 20VC with Harry Stebbings
- ▶ 45:10 Jonathan Ross Previously, for, it was, it was famously reported that, uh, OpenAI's GPT-IV had, I bel-, it started off with something like, 16 experts, and they got it down to eight, I forget the numbers, but it, like, started off larger, and they shrunk…
George Sivulka, Co-Founder & CEO @Hebbia: The Future of Foundation Models | E1250 · 20VC with Harry Stebbings
- ▶ 39:13 George Sivulka So then GPT-IV was released, I think, like a few weeks later.
Sam Altman: What Startups Will be Steamrolled by OpenAI & Where is Opportunity | E1223 · 20VC with Harry Stebbings
- ▶ 3:04 Sam Altman If, on the other hand, you build a company that benefits from the model getting better and better, if, you know, an oracle told you today that, oh, four was gonna be just absolutely incredible and do all of these things that right now feel…
- ▶ 5:39 Sam Altman And so, and they were doing these things where we could already see GPT four coming and we're like, man, it's going to be so good.
- ▶ 15:48 Sam Altman It's really easy to copy something you know works, and one of the reasons that people don't talk about about why it's so easy is you have the conviction to know it's possible, and so after, after a research lab does something, even if you…
- ▶ 25:21 Sam Altman Well, when we started working on GPT-IV, there were some issues that caused us a lot of consternation that we really didn't know how to solve.
Bret Taylor: Why Pre-Training is for Morons & Companies Will Build Their Own Software | E1209 · 20VC with Harry Stebbings
- ▶ 18:54 Bret Taylor They see these amazing models like GPT for available.
- ▶ 23:27 Bret Taylor Similarly, when GPT four came out, I, I haven't done the math on it, but it certainly had a meaningful lead for quite a while.
- ▶ 25:04 Harry Stebbings Can, can I ask, can you continue to have step function changes with every model in terms of GPT-III to GPT-IV? 2 times in the scene
- ▶ 47:12 Bret Taylor I think with GBT four, we crossed, uh, that quality threshold of effectiveness with conversational AI, meaning you can now have a conversation with a computer and it actually works.
Zico Kolter: OpenAI's Newest Board Member on The Biggest Questions and Concerns in AI Safety | E1197 · 20VC with Harry Stebbings
- ▶ 46:53 Zico Kolter Number two, if you look at the current best models that there are right now, so things like GPT-IV, Claude, 3.5, um, Gemini, things like this, I would not currently be all that nervous about having an open source model that was as capable…
Arvind Narayanan: AI Scaling Myths, The Core Bottlenecks in AI Today & The Future of Models | E1195 · 20VC with Harry Stebbings
- ▶ 4:30 Arvind Narayanan So I, in my, in my view, at least the biggest thing that changed between GPT, 3.5 and GPT four was the size of the model. 2 times in the scene
- ▶ 18:44 Arvind Narayanan So when GPT-IV came out and OpenAI claimed that it passed the bar exam and the medical licensing exam, uh, people were very excited slash, uh, scared about what this means for doctors and lawyers, and the answer turned out to be…
- ▶ 44:24 Arvind Narayanan So I think, like a lot of people, I was fooled by how quickly after GPT-III.V, GPT-IV came out. 5 times in the scene
Aidan Gomez: What No One Understands About Foundation Models | E1191 · 20VC with Harry Stebbings
- ▶ 5:26 Aidan Gomez If you look at the past, let's say like a year and a half, so between, I guess by now it would be like between ChatGPT coming out or, or GPT-IV coming out, uh, and now. 3 times in the scene
Ethan Mollick: Why OpenAl Abandons Products, The Biggest Opportunities They Have Not Taken | E1184 · 20VC with Harry Stebbings
- ▶ 3:13 Ethan Mollick We now have an open source GPT four capable model 2 times in the scene
- ▶ 26:11 Ethan Mollick When you look at companies, they have, um, so first of all, a lot of them don't even allow access to GPT-IV because the regulatory environment's unclear.
- ▶ 48:39 Ethan Mollick I mean, I hope so, but let's just, the first randomized control trial we have, I have some of my colleagues at Wharton, um, was giving GPT-IV people for math tutoring in Turkey. 2 times in the scene
Matt Clifford: The Bull & Bear Case for China's Ability to Challenge the US' AI Capabilities | E1172 · 20VC with Harry Stebbings
- ▶ 11:32 Matt Clifford So actually another way of thinking about it is one thing you've seen over the last couple of years since GPT-IV was released is actually a convergence of capabilities. 3 times in the scene
- ▶ 26:07 Matt Clifford If they're, like, wildly better, if it's as big a gap between four and five as it was between three and four, that is a very different world from if it really feels like we're eking out these tiny improvements despite having invested truly… 2 times in the scene
- ▶ 40:06 Matt Clifford Um, you know, I, I don't think GPT-IV is, um.
David Luan: Why Nvidia Will Enter the Model Space & Models Will Enter the Chip Space | E1169 · 20VC with Harry Stebbings
- ▶ 9:17 David Luan Historically, if you go look all the way back to GPT-II to GPT-III to IV, et cetera, um, what, the way it's worked is that the, um, you know, every, let's say, just to use a reductionist analogy, every incremental GPU you throw at the…
- ▶ 47:24 David Luan Right now, right, imagine you're, you're, you're very large company X, right, and you need capability Y, and then you've got the base model over here, right, that's pretty darn smart, GPT-FOR, or, or Gemini, or whatever, right, um, and…
Alex Wang: Why Data Not Compute is the Bottleneck to Foundation Model Performance | E1164 · 20VC with Harry Stebbings
- ▶ 1:13 Alex Wang I mean, I think there's been, um, this is especially coming up now, where GPT-IV, uh, you know, OpenAI has had GPT-IV since fall of twenty-twenty-two. 7 times in the scene
- ▶ 8:29 Alex Wang Um, the GPT-IV was trained on an internet data set that was less than
- ▶ 11:23 Alex Wang And we need, uh, we need the means of production of new frontier data to get, to get us from, you know, GPT-IV to GPT-IV. 4 times in the scene
- ▶ 37:24 Alex Wang So, um, when, you know, when OpenAI first produced GPT-IV in the lab, uh, you know, China were nowhere near that.
Reid Hoffman: The Future of TikTok and The Inflection AI Deal | E1163 · 20VC with Harry Stebbings
- ▶ 20:16 unnamed speaker So it isn't just, you know, uh, put in data, right, you know, put in compute, press button, go away for six months, come back, and go, ha ha, I've got GBD-IV.
- ▶ 25:20 Reid Hoffman Because part of the idea with inflection, and still inflection, 2.5 is, you know, a unique GBD-IV class model that is the best one that prioritizes emotional intelligence along with, you know, kind of analytic intelligence.
- ▶ 55:01 Reid Hoffman Don't try to try to say, we are gonna have our own GPT-IV model, or whatever the, the current time slice of it, and that will be our own, you know,
Aravind Srinivas:Will Foundation Models Commoditise & Diminishing Returns in Model Performance|E1161 · 20VC with Harry Stebbings
- ▶ 8:53 Aravind Srinivas Uh, they even wrote a paper on it saying they trained their own foundation model, and that model, uh, is, is beaten convincingly by, like, a GPT-IV on all the finance benchmarks.
- ▶ 21:19 unnamed speaker I think GPT four quality models are not yet commoditized. 4 times in the scene
- ▶ 21:53 unnamed speaker Like, I wouldn't say GPT four O is like a lot smarter than GPT four turbo.
- ▶ 59:33 Aravind Srinivas I think that, that, that's amazing, um, vision to have, and, um, that's the sort of thing that doesn't work today, like, like GPT-IV cannot, cannot do it yet.
Aaron Levie: How the Business Model of SaaS Changes Forever & Startups vs Incumbents:Who Wins?|E1155 · 20VC with Harry Stebbings
- ▶ 5:42 Aaron Levie Versus GPT-IV versus, you know, Claude. 2 times in the scene
Sam Altman & Brad Lightcap: Which Companies Will Be Steamrolled by OpenAI? | E1140 · 20VC with Harry Stebbings
- ▶ 21:54 Sam Altman If you're building something on GPT-IV that a reasonable observer would say, if GPT-V is as much better 2 times in the scene
- ▶ 28:51 Sam Altman With GPT-IV, people do use it just in very, to help them do science, just in extremely primitive and limited ways.
- ▶ 36:31 Brad Lightcap A lot of companies think it's static, so a lot of companies think GPT-IV is the best the models will ever get.
Roundtable #7: Spotify, Adobe and Linkedin on How AI Changes The Future of Product & Design | E1097 · 20VC with Harry Stebbings
- ▶ 0:34 Gustav Söderström They need to understand GPT-IV as well as they understand the user.
- ▶ 9:58 Gustav Söderström They need to understand basically GPT-IV as well as they understand a user, what, what, what, uh, what it needs.
- ▶ 26:10 Gustav Söderström There was this, um, this, um, paper called mad prompt that came out recently where, you know, GPT for like a general model. 2 times in the scene
The Ultimate AI Roundtable: What Happens Now in AI, Why Google are Vulnerable | E1085 · 20VC with Harry Stebbings
- ▶ 3:07 Emad Mostaque You see the quality of GPT-III versus GPT-IV.
- ▶ 8:42 Richard Socher I predicted that we'll have a GPT-IV equivalent model 2 times in the scene
Jeff Seibert: Why OpenAI Will Become an Infrastructure Play | E1085 · 20VC with Harry Stebbings
- ▶ 13:03 Jeff Seibert And so then last year, literally all of a sudden GPT three comes out, chat GPT comes out, GPT four, and we started experimenting and we're like, Whoa, hold on.
Des Traynor: How to Survive and Thrive in a World of OpenAI | E1082 · 20VC with Harry Stebbings
- ▶ 17:43 Des Traynor It uses GPT-IV by OpenAI. 2 times in the scene
- ▶ 20:49 Des Traynor Even within OpenAI's ecosystem, you can use GPT-IV turbo, GPT-IV, GPT-V turbo turbo, you know, and like, they cost different amounts, and like, they're good at their, like, um, some of them are faster than others, some of them are more…
- ▶ 20:49 Des Traynor Even within OpenAI's ecosystem, you can use GPT-IV turbo, GPT-IV, GPT-V turbo turbo, you know, and like, they cost different amounts, and like, they're good at their, like, um, some of them are faster than others, some of them are more…
Guillermo Rauch: Why Great Companies are Defined by How Many Things They Say No To | E1069 · 20VC with Harry Stebbings
- ▶ 1:00:56 Guillermo Rauch Like if I use a lot of GPT-IV, it doesn't matter that you're charging me like 20, 30 bucks a month.
- ▶ 1:04:02 Guillermo Rauch Right now, Llama is not as good as GPT-IV. 2 times in the scene
Christian Kleinerman: Do OpenAI and Anthropic Have a Sustaining Moat? Who Wins the AI Wars? | E1063 · 20VC with Harry Stebbings
- ▶ 19:01 Christian Kleinerman There are some very shallow wrappers on top of a GPT-IV. 3 times in the scene
20VC: The Biggest AI Leaders on What Matters More; Model Size or Data Size & Where Does The Value in AI Accrue; to Startups or to Incumbents
- ▶ 13:36 Douwe Kiela GPT-IV kind of hit this ceiling.
- ▶ 18:46 Douwe Kiela But I think if you look at the secret sauce to a lot of these other models, like why is GPT-IV so awesome? 3 times in the scene
Nikhil Basu Trivedi: Why 99% of AI Investments Will Go Bust | E1057 · 20VC with Harry Stebbings
- ▶ 59:18 Nikhil Basu Trivedi And there's also just the reality that as ChatGPT itself gets better in leveraging GPT-IV and other 2 times in the scene
Howie Liu: Decoding Airtable's $11B Valuation; The Impending AI Revolution in Enterprise | E1053 · 20VC with Harry Stebbings
Richard Socher: The 3 Biggest Barriers to Building AGI; AI Startups vs Incumbents | E1050 · 20VC with Harry Stebbings
- ▶ 28:00 Richard Socher I predicted that we'll have a GBD four equivalent model before the end of the year. 2 times in the scene
Douwe Kiela: Why Data Size Matters More Than Model Size; Why Open Source Isn't Going to Win | E1032 · 20VC with Harry Stebbings
- ▶ 16:29 Douwe Kiela Um, and, uh, so that, that GPT-IV kind of hit this ceiling.
- ▶ 19:00 Douwe Kiela But I think if you look at the secret sauce to a lot of these other models, like why is GPT-IV so awesome? 3 times in the scene
- ▶ 25:24 Douwe Kiela Uh, so what we've seen people do in the field now is they're using GPD for, to evaluate the quality of other language models. 2 times in the scene
- ▶ 30:42 Douwe Kiela Uh, so these are GPT four and, and tropic models and things like that, that are just much better than everything else, but also much more expensive and much bigger than everything else.
- ▶ 42:02 Douwe Kiela And, and a lot of work is right now actually going into finding the direct use cases for this technology, because people are now starting to think that they can use GPT, GPT-IV for anything.
Alex Lebrun: Why the EU's AI Regulation is a Disaster; How Zuck Prepares for Meetings | E1027 · 20VC with Harry Stebbings
- ▶ 10:43 Alex Lebrun Um, so for instance, ChatGPT, you know, is based on GPT-IV, which is based on GPT-III, was released three years ago.
- ▶ 19:53 Alex Lebrun Uh, with only 1000, uh, question and answer examples, so very, very small data sets, um, they get something for, use for fine tuning, so the second stage, they get something that performs better than GPT-III, and almost at the level of…
Tomer Cohen: Why LinkedIn Stories Failed; How LinkedIn's Feed Was Born; AI Startups | E1019 · 20VC with Harry Stebbings
- ▶ 1:26:19 Tomer Cohen There is now like auto GPT that takes the GPT-IV and builds, uh, kind of, uh, overarching, um, kind of, uh, objective function above it that kind of breaks it down to multiple subtasks.
Emad Mostaque: These 5 Companies Will Win the AI War; Why We Need National Data Sets | E1015 · 20VC with Harry Stebbings
- ▶ 4:09 Emad Mostaque What if you had a thousand GPT-IVs organizing all that knowledge, and then make it available to everyone, so you can see the exact potential mechanisms that way which MS works, and all the potential food, 2 times in the scene
- ▶ 14:11 Emad Mostaque If you have an option of having a stable chat model, uh, which will be announced in the future, um, that you own completely, trained in your own cloud, or on-prem, or on-device, and then also using GPT-IV, that's the best of both worlds.
- ▶ 23:45 Emad Mostaque India, all of the outsourcing jobs in programming will go, because GPT-IV can go level three Google programmer exam and pass it.
- ▶ 27:32 Emad Mostaque and then it took six months to turn GPT-IV into ChatGPT-IV.
- ▶ 30:28 Emad Mostaque GPT-IV can write you an article about any news that's happening in a way that's customized to you and your context and everything like that.
- ▶ 36:46 Emad Mostaque And I will tell you that GPT-IV is great, or Cohere is great, or all this stuff.
- ▶ 37:46 Emad Mostaque You know, you see the quality of GPT-III versus GPT-IV. 2 times in the scene
- ▶ 51:17 Emad Mostaque Versus GPT-IV? 2 times in the scene
- ▶ 53:34 Emad Mostaque And you think about certain industries and how they'll be affected by having the ability to have a thousand GPT-IVs working together. 2 times in the scene
- ▶ 58:23 Emad Mostaque Um, because I already use GPT-IV as a therapist and things like that.
- ▶ 1:06:25 Emad Mostaque Because anyone that you've talked to is like, how hard was it to implement GPT-IV?
Clem Delangue: The Ultimate Guide to Investing in AI; Elon's Threat to Sue OpenAI | E1013 · 20VC with Harry Stebbings
- ▶ 19:35 Clément Delangue I don't know if you've seen, but, uh, Elon Musk tweeted that he's, he's gonna sue OpenAI for using, uh, tweets in their training for, for GPT-IV.
- ▶ 52:49 Clément Delangue Like all companies will have their, their chat GPT or their GPT-IV.
Vince Hankes: Why We Put $300M into OpenAI; Sam Altman's Pitch; Lessons from Josh Kushner | E1009 · 20VC with Harry Stebbings
- ▶ 14:58 Vince Hankes And ultimately, I mean, maybe in the classic Sam kind of way, the way we kicked off the round was he did almost like a closed demo with lots of investors on a couple of calls of the technology they're working on and ultimately in GPT-IV.
- ▶ 34:26 Vince Hankes You know, I don't know all the things that they're doing internally, but, you know, as an example on GPT four, we saw the demo in the fall. 2 times in the scene
Tomasz Tunguz: How I Raised $230M; ChatGPT vs. Google; How LLMs Work; Trump vs DeSantis | E1004 · 20VC with Harry Stebbings
- ▶ 47:16 Tomasz Tunguz I think it's part of it, but, you know, the cost per query, so you think about, like, the cost to produce a GPT-IV query versus the cost to produce a Google query, I bet it's like a hundred or a thousand or 10,000 times different.
Ariel Cohen: The Death of Salesforce; How OpenAI is Changing the Travel Industry | 20VC #975 · 20VC with Harry Stebbings
- ▶ 41:11 Ariel Cohen Uh, it's like they infused it with so much data, and that by the way, GPT-III, think about GPT-IV and how much more data is there.