GPT-5, every mention
79 scenes, the whole family · ← back to GPT-5
tap a year for its mentions
every year anyone Shawn Wang 29Alessio Fanelli 8Quentin Anthony 6Josh McGrath 6Alex Lupsasca 6Will Bryk 4Greg Brockman 4Kevin Weil 3Dylan Patel 3Bill Chen 3
Verbatim, from the transcripts: the passages where GPT-5 comes up
The Inference Frontier: from 100 to 10,000 tokens per second — Sean Lie, Cerebras CTO
Simulating Humanity: from Generative Agents to 8 Billion Digital Twins — Joon Sung Park, Simile AI
- ▶ 27:14 Alessio Fanelli Basically, if I was to do the same thing that you described with, say, your favorite LLM, Opus, GPT-Five-Six, have some agent to map out these things.
Next 100x in AI: Inference, Networking, & Self-Optimizing Models — Philip Kiely & Ali Taha, Baseten
- ▶ 1:08:46 Alessio Fanelli When you look at the other side, like, how long have we been on the GPT-V generation, right?
- ▶ 1:14:45 Ali Taha And it's one of those spaces where the, the open source models are, like with LLMs, we see Kimi K-III is almost comparable to, you know, Mythos or like GPT-Five.
Podcast Crossover: AIE, AGI, frontier lab strategy with @matthew_berman and @swyxtv
- ▶ 11:10 Matthew Berman So GPT-Five.
- ▶ 20:23 Matthew Berman Even from the previous generation Opus, uh, GPT-Five.
Satya Nadella on AI: @NoPriorsPodcast x Latent Space Crossover Special at Microsoft Build 2026
- ▶ 4:46 Satya Nadella We used, whatever, GPT-Five-Five, right?
The Agent-Native Cloud: 3M Users, 100K Signups/Wk, Data Centers, & Death PRs — Jake Cooper, Railway
- ▶ 1:16:07 Shawn Wang Uh, and this basically GPT-Five is just routing and flagging, you know, through, through different models.
🔬How GPT‑5 derived new results in theoretical physics and quantum gravity — Alex Lupsasca, OpenAI
- ▶ 2:32 Alex Lupsasca Then when GPT-V came out. 2 times in the scene
- ▶ 1:11:17 Alex Lupsasca And then GPT-Five came out six months later and 3 times in the scene
- ▶ 1:30:51 Alex Lupsasca And then GPT five was able to do some of
AI-Native Engineering: 100% adoption, 5x search throughput, unlimited tokens — Mikhail Parakhin
- ▶ 1:04:42 Mikhail Parakhin I mean, not a, not a panacea, not, not really, uh, I wouldn't say that it's frontier model in the sense of it's not going to suddenly compete with, uh, GPT-Five.
Extreme Harness Engineering: 1M LOC, 1B toks/day, 0% human code or review — Ryan Lopopolo, OpenAI
- ▶ 5:38 Ryan Lopopolo But yeah, so onward to GPT-V, five, one, five, two, five, three, five, four, to go through all these model generations and see their kind of quirks and different working styles also meant we had to adapt the code base to
⚡️ Reverse Engineering OpenAI's Training Data — Pratyush Maini, Datology
- ▶ 8:40 Pratyush Maini And so, my first thought was, is this something that happened always, or is it just like a GPT-V phenomenon? 2 times in the scene
⚡️ Prism: OpenAI's LaTeX "Cursor for Scientists" — Kevin Weil & Victor Powell, OpenAI for Science
- ▶ 6:49 Kevin Weil You know, you can go through, and you can change it, and then you compile that into, into the PDF itself, but here, I can say, um, this is where, at the bottom, you can use the, the AI, you can use GPT-V. 3 times in the scene
Brex’s AI Hail Mary — With CTO James Reggio (acquired for $5B by Capital One!)
- ▶ 49:16 James Reggio GPT-V thinks Brex does and how it thinks our business operates is actually quite different from what our business offers today or how our product works. 2 times in the scene
Artificial Analysis: The Independent LLM Analysis House — with George Cameron and Micah Hill-Smith
- ▶ 38:52 Shawn Wang Look at GPT five is this big circle.
- ▶ 1:06:07 Shawn Wang And also how does that laid out in over the course of the year where we have things like gypsy five, which is a model router? 4 times in the scene
[State of MechInterp] SAEs in Production, Circuit Tracing, AI4Science, "Pragmatic" Interp — Goodfire
- ▶ 10:09 Mark Bissell It's, it's the equivalent of using GPT-V as a judge, but it's, you know, like, 500 times cheaper.
[State of Post-Training] From GPT-4.1 to 5.1: RLVR, Agent & Token Efficiency — Josh McGrath, OpenAI
- ▶ 6:13 Josh McGrath You know, I think if you look at like deep research, the original one and GPT-Five thinking on like high reasoning today, I think you'll see that like eventually the models all sort of converge in their, their capabilities. 3 times in the scene
- ▶ 13:58 Josh McGrath Yeah, and so you can see, like, from five to 5.1, our, our overall evals, you know, we, we bumped some. 3 times in the scene
[State of RL/Reasoning] IMO/IOI Gold, OpenAI o3/GPT-5, and Cursor Composer — Ashvin Nair, Cursor
- ▶ 17:33 unnamed speaker And there, we merged the four O and O-one O-three line into five, and now we're splitting it out into five and five codecs again. 2 times in the scene
- ▶ 17:33 unnamed speaker And there, we merged the four O and O-one O-three line into five, and now we're splitting it out into five and five codecs again.
⚡️GPT5-Codex-Max: Training Agents with Personality, Tools & Trust — Brian Fioca + Bill Chen, OpenAI
- ▶ 3:02 Brian Fioca Um, so I got to, I mean, this sounds weird to say, but I, I was lucky enough to be really close to the training team while GP five was training.
- ▶ 5:05 unnamed speaker Well, the news-wise, I think the communication where the Five Codex just came out was, well, this is the model trade for our products, not necessarily your choice, right? 2 times in the scene
- ▶ 6:35 Bill Chen And that's sort of what Brian was referring to when he's talked about GPT-Five's, uh, tools. 3 times in the scene
- ▶ 10:39 Brian Fioca With the five series, because it's more general, and it's just about as good as coding as codex for a lot of things.
The Great Evals Debate — Ankur Goyal & Malte Ubl
- ▶ 30:05 Ankur Goyal And, you know, GBT five is like, okay at this, but I want it to be like really great at this problem.
Anthropic, Glean & OpenRouter: How AI Moats Are Built with Deedy Das of Menlo Ventures
- ▶ 1:16:47 Alessio Fanelli So it's like, you know, can you run GPT-Five Pro over the next 1015 years?
Terminal-Bench 2.0: the most impt coding agent benchmark of 2025 gets a v2! Launch + Q&A w/ founders
- ▶ 6:04 Mike Merrill We see codecs with GPT-V ranks at the top of the benchmark in our testing.
How Zyphra went all-in on AMD + Why Devs feel faster with AI but are slower — with Quentin Anthony
- ▶ 23:58 unnamed speaker The GPT-Five to GPT-OSS handoff, for example, that could happen, where you have OSS on the device, and then it hands off to GPT-Five for more compute-intensive tasks. 4 times in the scene
- ▶ 34:29 Quentin Anthony Um, if I need something a bit more heady, like, um, writing me, for example, the, the setup for kernels, like we were talking about earlier, I might use a thinking model, like an extended thinking on GPT five or an O three kind of prompt. 6 times in the scene
⚡️Automating Scientific Discovery - Jessica Rumbelow, Leap Labs
- ▶ 17:21 Shawn Wang And I think like the newer models, like the GPC fives of the world are optimizing for tool use and kind of freezing everything else.
The Agents Economy Backbone - with Emily Glassberg Sands, Head of Data & AI at Stripe
- ▶ 1:03:42 Emily Glassberg Sands With, like, GPT-IV, like, it didn't work, and then you swap in GPT-V, and all of a sudden, it does.
⚡ Open Model Pretraining Masterclass — Elie Bakouch, HuggingFace SmolLM 3, FineWeb, FinePDF
- ▶ 9:43 Elie Bakouch I mean, I'm always curious to know, like, Claude, Claude Opus, like GPT-V, like what's the scale of those very big models?
Terminal-Bench: Pushing Claude Code, OpenAI Codex, Factory Droid, et al to the limits
Why RL Won — Kyle Corbitt, OpenPipe (acq. CoreWeave)
- ▶ 49:47 Shawn Wang They translate to a, a GPT five token.
DevDay 2025: Apps SDK, Agent Kit, MCP, Codex and why Prompting is More Important than Ever
- ▶ 30:12 Sherwin Wu But if you're a Plus or a Pro subscriber, you get routed to, uh, uh, GBT-Five, which is, I think what they- I think we also recently announced the partnership with Kakao.
- ▶ 40:15 Sherwin Wu I actually haven't tried this with, like, codec, GBD-V codecs.
⚡️Traversal: Causal ML and Reinforcement Learning
- ▶ 30:01 Shawn Wang Uh, and then like, obviously if you, if you like GPT-V, let's, you can talk about GPT-V as well. 6 times in the scene
Amp: The Emperor Has No Clothes
- ▶ 28:18 Quinn Slack You have GP five that came out and we're still figuring out how to use these things, but it would actually be an incredibly pessimistic outcome if all those smart people and all that money were not able to build anything that was better…
- ▶ 29:06 Alessio Fanelli Is the GPT-V codex only being available in codex making you nervous about future availability of, like, cutting-edge models, and, like, does that put more emphasis in, like, figuring out maybe, like, an open source strategy?
- ▶ 29:57 Alessio Fanelli Like when cursors switch from Sonnet to GPT-V as like the default model that was like, you know, 2 times in the scene
- ▶ 1:01:51 Alessio Fanelli And every model was failing until GPT-V Codex and Codex CLI was like the first time I got it in like one shot.
- ▶ 1:07:37 Alessio Fanelli So I built for like the GPT-Five post, it's like task manager that's like CLI first.
⚡️ Beyond Transformers with Power Retention
- ▶ 26:15 unnamed speaker Do you feel that from here, you know, what's next to like, start scaling this, you know, what, like what, how far do you think you are from being able to do GPT five on power attention, you know, kind of like when the frontier models do…
⚡️No, Don't Do Palantir for AI - Brendan Falk, Hercules (AUDIO FIXED)
- ▶ 26:26 Brendan Falk And candidly, I think that ChatGPT is going to come in and, and replace a lot of that because that's what they did in a GBT five demo.
Context Engineering for Agents - Lance Martin, LangChain
- ▶ 5:21 Shawn Wang Uh, we did our posts on Gpt.
- ▶ 1:02:48 Lance Martin We even have better results coming out soon with GPT-V.
⚡️Launching Ona: Coding Agent with Fully Sandboxed Cloud Environment
- ▶ 18:51 Shawn Wang Any discoveries from, like, using GPT-V versus Cloud-IV, you know, anything like that?
Greg Brockman on OpenAI's Road to AGI
- ▶ 0:22 unnamed speaker Congrats on GPT-Five, GPT-OSS, like all the stuff that's going on in OpenAI Lands. 3 times in the scene
- ▶ 1:27 unnamed speaker What was the path from there to GPT-V, where now it's like, kind of hidden from the user?
- ▶ 16:00 unnamed speaker We're in the, you know, multiple low double digit to high single digit range for, you know, GPT four, 4.5 and five, but we, you know, we're not confirming that, but like, you know, we're, we're, we're scaling there.
- ▶ 18:14 Greg Brockman A GPT-III or GPT-IV, not a GPT-V for sure, right?
- ▶ 19:33 unnamed speaker How would you characterize the beginning of the GPT-V era? 8 times in the scene
- ▶ 26:50 Greg Brockman GP five is very good at front end, it turns out.
- ▶ 36:13 Greg Brockman And GPT-V itself is extremely good at instruction following.
- ▶ 39:26 unnamed speaker One thing I wanted to touch on for the, I think the last kind of last couple of topics on GPT five, before we move to OSS, you've acknowledged that there's a router, which is really cool. 4 times in the scene
- ▶ 46:11 unnamed speaker Um, for the GPT-V release, I did this article called self improving coding agents. 3 times in the scene
- ▶ 59:55 Greg Brockman And even for something like GPT-V that, you know, we, we sort of had a lot of pressure to think about, okay, let's just like, you know, sort of do the grind of like, you know, here's feedback on problems that we have on the coding side.
The AI Agenda: GPT5 leaks and the business of AI News — Steph Palazzolo, The Information
- ▶ 26:39 Shawn Wang You know, uh, we're, we're recording this on with, uh, GPT-Five imminent. 9 times in the scene
- ▶ 1:06:06 Stephanie Palazzolo I think first, like, the first thing is like, I think this is what I'm most interested in, but, um, I did kind of touch on it with like our conversation about GPT-Five, but just like the trajectory of AI progress and kind of like, you know,
The RLVR Revolution — with Nathan Lambert (AI2, Interconnects.ai)
- ▶ 39:57 Nathan Lambert So it's like, but it's also the GPT-V idea, which is how do you get a model that just routes the question to the right?
⚡️ARC-AGI-3: The Interactive Reasoning Benchmark
- ▶ 28:10 Greg Kamradt I'm more confident to say that if you have a base LLM, like a GPT five, if, if GPT five is a base LLM without all the routing and COT and all that crazy stuff, uh, it's not gonna, it's, I'm pretty confident it's not gonna beat, um, arc two.
DeepSeek V3, SGLang, and the state of Open Model Inference in 2025 (Quantization, MoEs, Pricing)
- ▶ 15:11 unnamed speaker Like the, the failures of 3.5 Opus, the failures of GPC five, like, you know, it's a, it's a thing that people are sort of rumoring.
Beating Google at Search with Neural PageRank and $5M of H200s — with Will Bryk of Exa.ai
- ▶ 16:42 Will Bryk You could have a world, and we are going to have this world, where you have, like, GPT-V level systems and beyond that could, like, answer any complex request.
- ▶ 48:52 Will Bryk Specifically for search, I think it's interesting, it's different between, like, search and, like, GBP-V are different, because GBP-V might benefit from training on a lot of PhD notes, because 3 times in the scene
2024 Year in Review: The Big Scaling Debate, the Four Wars of AI, Top Themes and the Rise of Agents
- ▶ 16:52 Shawn Wang And what that means is basically like, it's very convenient that we're not getting GPT-Five, which would have been a larger pre-train.
- ▶ 1:23:09 Shawn Wang Um, a, uh, we also picked out, uh, that Sam was talking about GPT five soon. 2 times in the scene
The Winds of AI Winter (Q2 Four Wars of the AI Stack Recap)
- ▶ 42:11 unnamed speaker I think GPT-V or Gemini-III or whatever, uh, would be much more capable at this kind of stuff while it also serves our needs, uh, with like, with everyday things.
Supervise the Process of AI Research — with Jungwon Byun and Andreas Stuhlmüller of Elicit
- ▶ 29:01 Jungwon Byun What happens if GPT-V comes out? 2 times in the scene
The State of AI in production — with David Hsu of Retool
The "Normsky" architecture for AI coding agents — with Beyang Liu + Steve Yegge of SourceGraph
- ▶ 1:00:07 Shawn Wang So, five is, uh, you know, a hundred, a hundred people, uh, and if you just project it out, uh, nine is, um, every human on earth, and, uh, 10 is every human ever.
The State of Silicon and the GPU Poors - with Dylan Patel of SemiAnalysis
- ▶ 34:27 Dylan Patel Open source will, will match GPT-IV, but then it's like, what about GPT-IV vision, or what about, you know, uh, five and six, and, you know, all these kind of stuff, and like, Interact tool use, and Dolly, and like, that's the other thing,…
- ▶ 50:05 Dylan Patel Like, and, and like after GPT-V releases, if he goes to the market and says like, hey, I want to raise a hundred billion dollars at five hundred billion dollar valuation, I'm sure the market would give it to him, right? 2 times in the scene
Beating GPT-4 with Open Source Models - with Michael Royzen of Phind
- ▶ 37:38 Michael Royzen As was true last year and, and before, um, I think that GPT-IV, my hypothesis is that the jump from four to 4.5, or four to five, will be smaller than the jump from 2 times in the scene
The End of Finetuning — with Jeremy Howard of Fast.ai
- ▶ 1:18:41 Jeremy Howard Definitely is different to just feeling like it's probably on its way, you know, and yeah, whatever GPT-V looks like, um, I'm sure, I imagine I'll have the same visceral reaction, you know.