Glory Jain

Sora 2 is here.

wen GPT-5? In 10 minutes. openai.com/live

2025 is the year of agents. baby steps in that direction:

o3-mini is here! Together with @shengjia_zhao, @_kevinlu, @max_a_schwarzer, @ericmitchellai, @brian_zq, @sandersted and many others, we trained this efficient reasoning model, maximally compressing the intelligence from big brothers o1 / o3. The model is very good in hard math/coding/science questions with a fraction of cost and latency, defining new cost-efficient reasoning frontier. The model supports three different reasoning powers. Users can adjust the thinking time based on different use cases. The longer the model thinks, the better its capability is. With o3-mini-low, we drastically reduce latency compared to o1-mini, achieving GPT-4o level latency for response. You can apply early access to this model today to do safety-testing! openai.com/index/early-ac…

sora is launching today to all chatgpt pro and plus users! it's been a big effort to make this possible + i think the product is really fun & intuitive. my fav thing to do is generate fake historical found footage. video inpainting is also really strong. have fun!

so excited for Sora to launch today! seeing words in a textbox (a corgi with sunglasses surfing with the golden gate bridge in the background) come to life within seconds is nothing short of magic 🪄✨

🎄🎅starting tomorrow at 10 am pacific, we are doing 12 days of openai. each weekday, we will have a livestream with a launch or demo, some big ones and some stocking stuffers. we’ve got some great stuff to share, hope you enjoy! merry christmas.

completed 2 years today at OpenAI! grateful for the incredible people I get to work with and the chance to be part of 🤯research that can truly make a difference. excited for what's ahead on this journey 🚀

@TheRealAdamG @OpenAI congrats Adam!! so happy I get to work with you and learn from the best 🚀🚀

Today, November 1st, is day one of year four for me at OpenAI. I joined as the first salesperson on the GTM team back in November 2021. I often get asked about how I landed at @OpenAI so early on. Below is a long form, but abbreviated, summary of the dots that connected that led me to OpenAI. The story begins during my time (2013-2016) at Pivotal - the company behind Cloud Foundry. However, it was my experiences with the Pivotal Labs division that set me on the path that led me to OpenAI. Pivotal Labs was a software development shop that practiced a sect of Agile development called "Extreme Programming" - ideally using Ruby on Rails, but they grew to like JavaScript too. Extreme Programming was anchored in pair programming and test-driven development. Picture two people working with two monitors, two keyboards, and two mice, but one shared computer. One person pilots while the other co-pilots. Think of an airplane cockpit with two yolks, but with one plane. The pilot was flying and the co-pilot was riding shotgun. I described it as "my wife and I writing an email to one of my daughter's teachers - someone was on the keyboard and the other was looking over their shoulder reading for flow, grammar, tone, missed words, etc. That approach delivered amazing code quality and culture, but it took a special type of developer to be able to engage directly with someone for hours on end. It was also tough to scale. But when it worked, it worked beautifully. Customers would come into our offices (SF, NYC, Boulder, Atlanta, etc.) to learn how to build products 'the Pivotal Way.' - while building production software. Clients ranged from startups like Twitter (venturebeat.com/business/messa…) to major banks, retailers, and automotive giants. While I left Pivotal to join the Azure team at Microsoft, you can never really take the "pivot" out of the employee -- I was a changed man from my experiences at Pivotal. I often described my time there as 'an MBA in all things that matter for the next 10 years in enterprise tech' — spanning everything from app development, DevOps, mobile, open source, containers, cloud, big and fast data, machine learning, GPUs, daily standups, retrospectives, and giving high-quality feedback. From the day OpenAI was founded I was drawn to the company. I remember reading articles in early 2016, reading about Dota and Rubix cubes -- through the Microsoft's investment in OpenAI in 2019. However, I was more of a fan than a future employee. When GPT-3 dropped and in 2020 it caught my attention as I was seeing the early sparks of something. I remember being drawn to open-endedness, things like "text to SQL" and this idea of "scaling laws". But I was still just an observer — a fan, an interested party. However, in 2021, OpenAI released the Codex model (openai.com/index/openai-c…) and helped Microsoft build the first GitHub Copilot. That was the moment for me - click, switch flipped. From the first time I saw the Codex launch video (staring @ilyasut, @gbd and @woj_zaremba - youtu.be/SGUCcjHTmGY), I knew that coding and software development was about to be transformed forever. Instead of two people pairing, it would be a person pairing with their uber smart AI. Devs wouldn't need to go search on Stack Overflow or Google for code snippets or guidance. Their AI would work with them through an issue or provide a code snippet (and so much more). I don't think we've fulfilled that vision yet, but we're moving fast on the path. I no longer doubt we'll get there. We will get there — it's more when, and what the user experience ultimately looks like and what's the role of the developer at that point? I don't even mean AGI. That's something else, but this is an important steppingstone. When I see what the new ChatGPT Canvas starts to unlock that future I saw in 2013 has started to become very real — where pairing with an AI-native assistant can give you true sustained lift. It's a privilege to work with some of the best people and rep some of the most amazing technology on a daily basis. Being afforded the opportunity to, at times, to work closely with people like @BorisMPower, @bradlightcap, @jasonkwon, and others during the early days of DALL-E 2 and the "low key research preview" that is ChatGPT have become core memories. If you view dog years in 7-years chunks, then AI years are 5-year chunks. So, it makes sense that I feel like I have been here a decade and half. October 2024 was a hoot. The rest of Q4 2024 is going to be fun AF. And for the record DALL-E 2 is still my favorite piece of OpenAI tech - it's where Broccoli Man was born. Thanks for indulging me as I reflected back. What a ride it's been, what a ride it will be.

The real size of things ~ A Thread🧵 1. If Saturn were as close to Earth as the Moon, this is what it would look like:

@amorriscode lovely meeting you too!

@_gloryjain nice meeting you today!

dev day year 2! from realtime API to model distillation, we keep shipping 🚀

🍓 there's a lot of o1 hype on my feed, so i'm worried that it might be setting the wrong expectations what o1 is: the first reasoning model that shines in really hard tasks, and it'll only get better. (i'm personally psyched about the model's potential & trajectory!) what o1 isn't (yet!): a miracle model that does everything better than previous models. you might be disappointed if this is your expectation for today's launch – but we're working to get there!

@edwinarbus maybe the hardest unsolved problems in artificial intelligence.... finally cracked....

hey o1, how many r's are in strawberry?

chatgpt 🤝 apple #wwdc

it’s here 🔥