Introducing ChatGPT Work, powered by Codex and GPT-5.6

0:00[music]

0:06Hi everyone. I am so excited to be here. Today, we are releasing our latest and most capable models. Terra, and Luna. Soul will be available over the next 24 hours to all our paid plans. And Terra and Luna are also coming to all our free users. Today, almost a billion users use ChatGPT every week. They use it for helping getting them to write, getting answers, do research, and all sorts of creative things. But what we've seen, and especially with Codex and the coding abilities, is that these models are capable of much more. We saw a transformation of software development, and shortly after, internally within OpenAI, everyone was using these models

0:57to do so much more than getting great answers. They were using these models to do great things. And you were expecting that we were going to release these models, and we are rolling them out over the next 24 hours. But what I'm really excited to be here today is we're not only releasing the models, we're also announcing three major product updates. The first one is

1:19ChatGPT Work. ChatGPT Work is your new partner for ambitious work. It works on web, it works on mobile on the go, and also on desktop. The models with GPT-5 6 Soul have really been a breakthrough in their ability to do things of complex nature for you. Understanding very complex content, staying on track for long horizon tasks, understanding spreadsheets, creating

1:45them, editing even websites. And we're going to be showing you first ChatGPT Work. We're also announcing an all brand new ChatGPT desktop app. It goes even further and allows ChatGPT to work on your local files, your apps, and your browser. Finally, we're also announcing hosted sites, which we're making available for all paid users. It allows you to use ChatGPT work to create and share those websites with your friends and colleagues. It is a truly magical experience, and I can't wait to show

2:20you. First, we're going to have a look at ChatGPT work. Jessica, Lauren, why don't you show us what they can do? Hi, I'm Jessica, engineer on ChatGPT work. And I'm Lauren, I'm on our strategic finance team. So, we're going to show you the new experience on web and mobile. So, let's take a look at mobile. So, here we have chat. Chat is the familiar thing you love you know and love. Um it's really great for quick answers, um search, and brainstorming ideas. But today, we're also introducing work. And work is where I go to have ChatGPT ChatGPT help me just get stuff done. Um so, take so

2:59Um so, for example, something that I did right before for to prepare for this launch is Hey ChatGPT, I would love to get an understanding of how people are using the ChatGPT work feature internally. Um look through Slack and employee feedback and find people in diverse roles um that have used the product in really interesting ways. I'm in San Francisco this week for the launch. I would love to meet up with a few people and really

3:28deep dive on their use cases in person. So, I can send this through mobile, but um these conversations also appear on web, and this is something that I actually did um earlier. We can take a look at the results, and ChatGPT work was able to pull from the sources that I asked for, uh find interesting people to talk to, and then schedule meetings with them while I'm in town. And one of the feedback posts that

3:53highlighted was actually Lawrence. Yeah, our finance team have become such power users of ChatGPT work. I'm really excited to show you how we've been using it, but I would love to hear how other teams have been using it, too. Yeah, we've talked to people throughout the company like recruiters that are using scheduled tasks to track interview progress and make sure that we're getting feedback timely and data science who are using our new visualize feature to just add context for quick ad hoc

4:18requests. But I think the finances experience was one of the ones that was most compelling to us. Yeah, I'm I'm excited to show it to you guys. So finance may not seem like the flashiest demo we could be leading this off with, but ChatGPT work has really enabled us to run with such lean and efficient teams and we wanted to show you guys a little bit of a peek behind the curtain at how we've been able to do that. So one of the critical roles of a finance team is being able to explain not only our recent trends, but what they mean for

4:45forecast in real time. This takes a couple minutes to run, so we did run this earlier today, but let's say we just ran just closed June we beat our forecast by $2 million. And I do need to caveat these are demo numbers. Tibo look like the numbers that you shared yesterday, so I'm not sharing the real

5:02numbers here. Tibo obviously beats his forecast by more than $2 million. These are These are demo numbers, but a very real a very real workflow. So we ran this earlier today. This used to take so much manual work. We would have to reconcile multiple systems, our Excel forecast model across multiple cases. And now in one pass ChatGPT can run the variance analysis for me. So we can see why we beat our

5:24forecast and where we still have risk. I can even have ChatGPT work go in and propose an updated forecast case for us. So we had it do this earlier we had it run it yesterday. It went through and updated our Excel model. Let's see. Perfect. Yep, we've got our updated revenue in here, our updated forecast. But I can't just walk Tibo and Jessica and our business partners, through an Excel model. So, we also had it put together a PowerPoint presentation that we can use. We've got our latest forecast. We've got our drivers bridge on what's changing in our outlook. Um, and ChatGPT work will meet you where you

6:00are. It's obviously it's great at Excel and PowerPoint, but our finance team have become such big proponents of sites. Um, it's such a flexible interface to be able to do custom custom dashboards, um, and really just like storytell around your analysis. So, I also asked ChatGPT work to go in and make us a site with the same analysis that we can share out. We've got our updated forecast. We've got our drivers bridge. This is looking good. It's good

6:25to go. Um, all right. So, we've run our variance analysis. Our forecast is updated. We've got our overviews back for business partners. Let's go ahead and have ChatGPT work send this to Tibo. Hey chat, can you send, uh, the site link over to Tibo on Slack? Thank you. And we'll send that off. I hope I get that soon.

6:45Yeah. It is such a delight to receive these highly interactive and compelling websites. Uh, in ChatGPT work, we've seen it perform very similar things across different use cases. Jessica was using it to understand user feedback in preparing ahead of the launch, even scheduling interesting chats with her colleagues in one-on-one through calendar. And that was Jessica, but Lauren used it for a very different kind of purpose across finance and very complex workflows where precision and

7:15correctness is extremely important. ChatGPT work was able to understand complex financial data and represent it in the right format at the right time just with a few steers. Now, I'm excited for us to show you the all new ChatGPT desktop app. Andrew, Ed, show us that what we can do with it. Yeah, thanks, Tibo. This is the new ChatGPT desktop app. So, everything that you saw on web, it's available here,

7:42plus a lot more on your local machine. So, local files, your browser tabs, even other apps on your computer, they're all available now to ChatGPT. So, let's take a quick look. Um I'm going to start with this spreadsheet right here that my colleague Nick sent me. Um it's an export of user tickets from our ticketing system. And it's it's pretty large. It's got a ton of columns. It would take a while to get through this and kind of tease out what the themes are and what we need to pay

8:07attention to. A lot of feedback. It's It's a lot of feedback. Um we're very used to this. Uh but instead, what we're going to do is um we're just going to drag it in. We're going to say, "Hey ChatGPT, can you make a quick interactive visualization of this feedback so that we can sort of get the themes and the action items and the criticality ranked and sort through it

8:31all a little bit better?" So, chat's going to go and take this spreadsheet, synthesize it the content, and then make a visualization. It'll take a second, so I'm going to walk you through something that I'm doing for next week's launch. So, I've got this folder over here, and this has a variety of material from various teams that have been sent to our our team here. Um this is a launch readiness PDF. And it's got a lot of stuff in it. Um then we have over here we've got um UXR. They did a bunch of interviews um with users on our early products. Um we've got a security

9:08review. This has all of our pen test results and even some of the compliance controls we need to meet. So, this folder's got a ton of stuff from various teams. They've just been kind of sending it to us um in prep for this launch. And I need to brief my team on Monday on the state of the launch, right? And so, what I'm going What I What I did this morning is I came in and I said, "Can you please look at the materials in this folder that everybody sent me? Um I also have three open tabs in Chrome that have content for this launch. Go

9:35look at it and make me a presentation that I can give to my team in the template that we use all the time at the company. Um and so it worked for about 90 seconds here, a little bit less. And at the end of this, um looked through my desktop, looked in Chrome at the tabs that were open, um and it produced a fully ready slide deck for me to present to the team about this

9:57launch. This is one thing that GPT-56 sole is really incredible at. It It went through all of that content in 90 seconds, synthesized it all, adhered to the template, and created created like a slide. You can just share that directly with the team. You don't have to do anything else. Exactly. And it used um it used its memory as well, cuz I've been, you know, chatting a a lot about this launch. And so it's got some of the concerns that I have that haven't actually been shared via Slack or email, too. Um But I know that I know that you were curious about the next World Cup game,

10:27and you were kind of busy um prepping for this live stream. So, uh we've got this we've got this really great feature in the app, um to do quick searches and anything that you want to know really quickly and get instant results. My friend Timothy wants to know when the World Cup uh game is so that he can

10:45watch it. Uh when is that? I'm actually more interested in the Belgium game from the other day. I know. I know. Um and so this is the ChatGPT search experience that everyone's come to know and love. It's almost instant results. It's um got rich search search widgets here, and um you know, anything you might want. But um

11:06back to work. Uh I've got this notes app situation that I don't think I'm unique in in that my Apple notes is just kind of a brain dump, and it's all over the place, and there's no folders, there's no organization. And um in a previous life, I would just kind of leave it this way and declare bankruptcy, and uh it'd be sort of a mess. But with the new ChatGPT desktop app, it actually has access to the other

11:28apps on my Mac. Organize my notes app. It's a It's a whole thing right now. Like make some folders, move the notes around, use your best judgment. I don't care. I just need to get out of this mess. So, what's going to happen here is ChatGPT with the incredible computer use that's built in and with the advances in the new model, it's going to get its own cursor. It's going to start operating Apple Notes in

11:57the background. Um you can see here. This is not my cursor. This is ChatGPT's cursor. Um there it goes. There it goes. There it goes. It's going. It's It's moving notes around. I can do my thing over here. I can check the Belgium game and when that is. Um and it's just going to go to town and make some folders and drag some notes around. Um we don't have to watch this whole thing um because I'm really actually quite curious of what Ed and team have

12:22been cooking. Can you show the visualization first second you started it? So, we were curious about all of this feedback that we had. Yeah, so here's the visualization made from the spreadsheet. We've got um it's like rich, it's on demand, uh it uses the theme that we're that we're using in the app. It kind of went down and ranked like, "Hey, here are the buckets. Here's how important they are. It's interactive. It's not static. So, I can, you know,

12:49change views. I can ask it for changes. And if I want to share this with my team because the spreadsheet's hard to read, in one click, I can just publish that to the site. That's a good idea. they get this exact same interactive visualization. Over to Ed. There are a few more examples, I think,

13:06right, as well? Yeah, there there That's a good call out here. This This is a set that I ran last night just kind of playing around with it. The These things are stunning. I mean, the the We didn't write any code to make these. The model is writing these on demand. The The ones are incredibly immersive

13:29and interactive. uses it as part of its answer, so it can sort of like model it like whether it answers in text or answers just like with a little visual like on the fly? Yeah, you you can So, you can ask explicitly for a chart or a visualization or sometimes if you ask to, you know, get educated. If you're like, "Hey, teach me about how this thing works." It'll come back with this interactive demo or if you say, "Help me sort through my inbox." It might even give you like a rich widget that shows

13:54you the maps and directions and stuff. It's It's It's awesome. Cool. Take it away. Hi, my name's Ed. I'm a designer on the team and I want to show you a little bit about how I use it and the team uses it in our day-to-day. So, you know, you walked through an example of creating assets for a launch using files on your computer. I've done the same thing here except I've used our new used our new sites feature. So, one thing I really want to highlight here is I kicked this prompt off this morning and it's looked through all of my connectors and plugins that we talked

14:25about earlier. I've connected my Slack, my Gmail, you know, everything that I use every day and it's gone through all of those sources and it's come back and turned it into this like super rich interactive website that I can go in. And the reason I want to show, you know, one big reason I want to show this is just because, you know, this is just one prompt kicked off and the visuals are really incredible. I mean, the the new 560 model

14:47Did you give it a Figma for this? No, no Figma. This was all all just the model. Yeah. Um and, you know, as a designer there are like little things you look out for like small motion, right? The typography, the um you know, the 3D motion and all that kind of thing. Um and it's it's always best when you give it a little bit of guidance and you, you know, you bring it along, but it's really outstanding out out of the box. Um and again, this is just like a really fun way that the teams have used to to use this feature across the company. But, you know, we're

15:14on live stream. We want to make it a little bit more exciting. So, what I annotate this um and let's change this header um to something a little bit more fun. Um, I think, you know, we saw some of the 3D visualizations that you showed, and I think the 3D, um, the 3D stuff is is super fun and interactive. So, um, hi

15:31ChatGPT, let's change this, uh, website. So, instead of having the static hero at the top with these planets spinning around, let's do an interactive game. So, I want to have like a little character that I can walk around with, and, you know, based on where I go, that will then take me to different parts, um, of the the live stream prep. So, there might be like launches and the run of show. Um, yeah. Take, you know, it might take a little bit of time, but but

15:53really, um, uh, go ahead and start that. So, while that's kicked off, I just want to show you a few other great examples. Um, you know, Lauren mentioned that their team use a lot of these for like dashboards and internal tools. That's definitely something that we've seen a lot, um, and you can see a lot of different examples from across the

16:11company here. A few just to call out. So, the web team, the openai.com team who build all of our amazing websites, they, um, instead of using like an Excel spreadsheet, they use this interactive tool where they can update it all collaboratively, they can share it with each other, and you can just hop in, you can see what's launched and when when it launched. You know, hover over, it's like so interactive and and again much nicer to look at rather than a spreadsheet. This just really

16:34illustrates it super well. Yeah, and it's really been, you know, night and day. Literally 3 months ago, spreadsheet. Today, these like, you know, it's it's really like transformed the way that the whole company works. Um, this is another great example. This is the ChatGPT images team. So, they've been collecting great examples of how ChatGPT images have been used across, um, all of our campaigns around the world. And when building sites as well and any kind of front end stuff, you know, we now have access to ChatGPT images as well, and it just also, you know, makes the whole experience like so much better. Um, as a designer, this is

17:03one thing that I quite like to do a lot. So, as part of the new launch, so you're not looking at, you know, ChatGPT or the ChatGPT desktop here, you're looking at, uh, an interactive prototype, um, that one of the designers on the team Tarik built and I've been working with him on. And it's a new way of prototyping our new model selector. Um you know, previously I might have just sent an

17:22image or or a or a design file to folks. Now I can show an interactive demo. I think I sent you this and you were like, "Whoa, this is like I can like click in and and test it out." It's so much better to illustrate like an idea and then you know, you just do it super quickly in a couple of minutes

17:34and then boom, you have it. Yeah. And then you share it with, you know, and then you just get the feedback you need, you incorporate it back. If you need to prototype a toggle or something. [laughter] Yeah. And the design team and like all teams across across Open AI really, you know, our work has changed so much and it's really been facilitated facilitated

17:49by these new models. And this is finally a bit of a fun one. So, you know, in the AI community, um a good test of kind of how good a model is at front end is this um uh this test where you kind of see if it can draw a pelican riding a tricycle um uh in an SVG. But Kian, one of the one of the folks on the team, he built this fun prototype where he built a 3D version. And again, this is just all 5.6 all um you know, cooking away, building out this this amazing prototype. And the really cool thing about Sides is they're

18:18collaborative, you can share them. And he shared this with people across the company and they've kind of added their own features. So, you know, the uh it can ride a pony, you know, it can ride all sorts of uh crazy things. It can ride another pelican. Um so, this is a like a hard example, but just like as a creative, it's just like so amazing how

18:35much it's expanded things. So, um I kicked off that prompt uh this morning. So, I'm just showing uh showing another example here of of an output. Oh, that's what we just Yeah, this is this was So, I kicked the same prompt off just before we started. Um and again, you can just see just how amazing it is out the park. So, it's the same website, um but now you can kind of go through and it's a bit more interactive and fun. Um but really just demonstrates that just in a few prompts now, kind of anyone can really build

19:01anything. It's really about raising the level of of ambition that you have, right? Totally. Um Thanks so much, Ed. Thanks so much, Andrew. Everything you saw here was really made possible for Jessica, Lauren across finance, getting help to schedule interviews, understanding things in the stress of a launch, and then here, just getting help to think ambitiously and executed on it really fast. All of that is made possible thanks to the model. And I'm here joined by Katie and Tejal to talk a little bit more about the research,

19:33which I'm tremendously excited about. And one thing really also to insist on is like it's not about taste, not just about front end. This model is our most capable coding model and it's also incredible at cyber, which we're going to talk about as well. Cool. I'm Katie. I'm a researcher at OpenAI, and it's pretty crazy to think that we've been here just for a year

19:51since the Codex launch. we were like precisely almost like Literally right here. Yeah. Hey, I'm Tejal. I'm also a researcher, and we're all so excited to share more about how we trained GPT-5.6. It's truly been a labor of love from our whole team across OpenAI. In 2024, we announced the reasoning paradigm, which was a reinforcement learning technique to tackle the hardest tasks. And since then, we've been scaling both reinforcement learning and pre-training, and we've seen exponential improvements in model capabilities as these two multiplied into together. And GPT-5.6 is the latest result of all of

20:25our research progress today. The 5.6 family brings these frontier capabilities to you. Soul is our most powerful model for the hardest agentic workflows. Terra is a faster model for everyday workflows, and Luna is our fastest and most affordable model for high-volume work. Already, Soul has been transforming our research program. As one example, 5.6 Soul actually

20:46autonomously post-trained Luna. This is the actual Codex problem so that we, a researcher on our team, used to have Soul kick off a post-training job for Luna. You can see here that, you know, it's like asking to find the training configs, find the right GPUs for the job, and then launch the script, and make sure that it works.

21:04It's a really short prompt. Yeah, it's pretty crazy that we've gotten far enough where you can have a fairly under specified prompt and then just give it to Codex and then just have it go and run the job. And you know, previously this is something that a team of senior researchers may have worked on at OpenAI and now it really feels like the automated researcher is pretty

21:21close. This isn't the only place we've seen internal acceleration. We're also seeing an increase in the number of pull requests per researcher, the number of experiments researchers can run so we can test our ideas and turn them into research findings. And actually Codex made every plot in this live stream and also helped us make the slide deck. So

21:38thank you 5.6. Across frontier evals, 5.6 soul is state of the art and this is something we're really excited for you to finally experience and see the speed up that we've already experienced internally. On terminal bench which is a test of coding performance, the model is state of the art. On browse comp which is an evaluation that sees if the model can locate hard to find information and the same with agent slots exam which is long

22:02horizon professional work. As one deep dive, 5.6 is particularly good at computer use so anything that involves navigating a browser, navigating apps on your desktop and that enables all sorts of digital work. It helps medical assistants navigate electronic health records, it helps data scientists analyze if a drug is effective, it can even help investment bankers create financial models. And you can also use this for personal use, it certainly helped me order food and all of this is something that when we ask experts in the field, you know, how they're using this model in their everyday workflows, they say it's better than anything else they've ever used,

22:35anything else on the market while being three times as fast. So we're really excited for you to finally try and feel how powerful this model is. You might hear the capabilities that Taj is talking about and that you saw demoed earlier and think that this is going to come at a very high price tag. But that's actually not the case with GPT

22:505.6 soul. So token efficiency has been a huge focus of our research program for several years and with every model we want to bring more intelligence for every token. And you can see this reflected on evals like deep sweep 1.1 where GPT 5.6 outperforms its competitors at less than half of the cost. But if you want the best intelligence money can buy, we're also announcing Ultra mode, which unleashes a whole team of agents to do work for you. So, in this example Eval, you can see here, you know, one agent does better with time, but as we add more agents, the model is able to do even better and faster. And

23:23that's because the agents are able to parallelize work like an experienced team. This is a capability folks have asked for so long that we bring to the Code Assist app, and I'm so excited that we're finally putting it out there as Ultra mode. One more improvement that we've shipped with GPT 5.6 is, you know, we may remember that in when we launched GPT 5.5, we had to ship it with a dev message saying, you know, don't talk too

23:47much about the goblins and the gremlins. And this is a result of reward hacking that occurred during the training of GPT 5.5, which has now been resolved in GPT 5.6. And now GPT 5.6 will only talk about goblins a tasteful amount when it's cute or appropriate. With these more capable models, safety and alignment are more important than

24:05ever. That's why we spent over 700,000 A100 equivalent hours of compute on model red teaming. We also spent 6 weeks safety training and testing this model, and also have incorporated multiple novel monitoring upgrades in our launch, including new classifiers and activation probes. Ultimately, we really believe in iterative deployment and putting our models in the hands of real people as quickly as possible so that we can learn from real-world usage outside of our labs. And that's why we have initiatives like Project Daybreak, where we give cybersecurity researchers access to our models earlier on. And with GPT 5.6 Soul, researchers have already found vulnerabilities in every major browser

24:42and database. Yeah. As part of the Daybreak umbrella, we started a Patch the Planet, which is an even broader initiative, uh where we work directly with open-source contributors and projects. And we also generate high-quality patches, for example, for Linux. Uh they accepted over half of our patches, which indicates that not only do we find critical vulnerabilities that weren't found for a long time, but also we were able to automatically patch them and

25:05greatly accelerating cyber defense. We're very proud of this model. We hope it helps you as much as it's helped us, and we're excited to see all that you build with 5.6 Soul. Thank you for sharing about the research and the model. What an incredible set of releases. We saw ChatGPT work capable of working on mobile and web. We also saw the all-new ChatGPT desktop app, and we saw all of the capabilities of GPT-5.6, Soul, Tera, and Luna, and a

25:37little bit about how they were trained. Next up, I'm going to be uh showing you a use case outside of the office that I'm tremendously excited about. Hiroki has been working with us for the last 6 months, and he's been using GPT-5.6 um to really do something unexpected and help him take care of his farm. Let's

25:58have a look at his story.

26:23[music]

26:28Hokkaido tractor [music] vegetable start engineer Codex greenhouse greenhouse [music] Codex window

27:27[music]

27:41[music]

27:45[music]

28:03[laughter]

28:24Hiroki, thank you so much for being here with us and traveling all the way from Hokkaido. I'm going to be using ChatGPT to live translate. Hey chat, I'm here with Hiroki-san, a dear and esteemed friend who traveled all the way from Japan. I want you to act as my live translator. He's going to speak Japanese, live translate to English to me, and I'm going to be speaking English, live translate back to

28:47Japanese. Got it. I'll do real-time Japanese-English both ways. Yeah. Whenever Hiroki-san starts, I'll do real-time translator, okay? Understood. Starting live translation now. Hiroki-san, transport us back to your farm to share the magic.

29:16Thank you so much for inviting me all the way to San Francisco today. I honestly can't believe I'm here right now. Right now I'm showing broccoli seeds, tilling the fields with the tractors, harvesting broccoli, and many other tasks all happening We'll start harvesting broccoli next week, so once this live event is over, I'll be heading back to Japan.

29:43We're just a small team. We're managing a very large area fields. So, improving efficiency is really important. To do that, we're making use of chat GPT to help us with our farm work. That's how we're moving things forward. Yes. You've really shared all of this from the very beginning. Why did you [clears throat] do that? And what were

30:07some of the reactions?

30:28How do you use AI? There are so many things where we don't have clear answers yet. Given that, the question is how we Excuse me, we still don't know yet. I think it's important to just keep putting things out there. What were some of the things that people shared with you?

31:08Lately I've been receiving messages from farmers all over the world. And I've also been contacted by people from Japan's Ministry of Agriculture, Forestry and Fisheries. All sorts of unexpected things have been happening. Wow, that's very unexpected. Um what [clears throat] what is what could possibly be the next challenge for you?

32:04Can you translate that to Japanese for me? Can you just restart that real quick? Hey Chad, I have Hiroki-san here, uh friend from Japan. I want you to act as my live translator, okay? Please, he's going to speak Japanese. Just live translate to English to me. Got it. Starting real-time Japanese to English translation now.

32:35within the team, I want us to make better use of AI. Right now, we're using our team's or our daily announcements and share updates on the We're trying to build systems so that our team members can work more comfortably with the idea that ChatGPT will support those systems. What you've been doing has been very advanced. What would you advise some non-technical people? Like, how can they

33:10get started?

33:27First of all, I started small. I wasn't an engineer myself. So, by tackling each thing I didn't understand, I asked ChatGPT about it. I gradually built systems that I can Now, I'm able to create and do things that are possible. So, I believe that starting small and building step by step is really important. Thank you so much, Hiroki-san, for sharing your journey with us and being

34:04live here today. Everything you just saw and the advice from Hiroki is profound. It is about taking small steps in raising the level of ambition you have in order to have ChatGPT work, get things done for you. What you saw and all of the product updates and the models are rolling out over the next 24 hours. Soul will be available for all our paid plans together with Chat GPT work, the desktop

34:36app, and sites as well. The Chat GPT desktop app is also available for free users. Soul is available for all our paid plans and Terran Luna for free users rolling out over the course of the next day. Thank you so much for tuning in. I can't wait to see what you do with it.