A floral poster in the style of Kehinde Wiley with a thoughtful theme and the word “OpenAI” in bold text
“o1 has set a new state-of-the-art on SciCode, beating Claude by a wide margin! Still *a lot* more to go here 🙂
“shipping a few new tools for developers today! from last devday to this one: *98% decrease in cost per token from GPT-4 to 4o mini *50x increase in token volume across our systems *excellent model intelligence progress *(and a little bit of drama along the way)” / X
Shh, ChatGPT. That’s a Secret. – The Atlantic
“The new Realtime API with web crawling is mind-blowing! Talk in realtime with any website. Powered by the OpenAI Realtime API and @firecrawl_dev 🔥 Check it out:
“Through April, 2023: “The release of ChatGPT led to a 2% drop in the number of jobs on [Upwork] and a 5.2% drop in monthly earnings. We find a similar reduction in the employment outcomes of freelancers offering design and image-editing services.” Large effects in a short time.
Exclusive: OpenAI to remove non-profit control and give Sam Altman equity | Reuters
Behind OpenAI’s Audacious Plan to Make A.I. Flow Like Electricity – The New York Times
“canvas is cool because oai is starting to explore novel interfaces. I like chatbots, but they’re the MVP of LLMs. Better interfaces will make LLMs much easier to use. See Cursor vs Copilot.” / X
My vision for the ultimate AGI interface is a blank canvas. The one that evolves, self-morphs over time with human preferences and invents novel ways of interacting with humans, redefining our relationship with AI technology and the entire Internet.
Examples of OpenAI Canvas:
1/ Do research in canvas! GPT4o with canvas can do research about art history, write a report, you can ask to verify its claims and then add citations and bibliography:
2/ Search for best restaurants and invite a friend to one of them in a written email
https://x.com/karinanguyen_/status/1841889814230061480
3/ GPT4o with canvas writes code in rust and then reviews code
https://x.com/karinanguyen_/status/1841889815689637979
4/ canvas model can browse for recipes and invent a new one based on the ingredients you have! https://x.com/karinanguyen_/status/1841889817522520246
“Sound On. This is what will come the ChatGPT desktop in a few months. Feel the AGI moment!
The new Realtime API with web crawling is mind-blowing! Talk in realtime with any website. Powered by the OpenAI Realtime API and
@firecrawl_dev Check it out:
“@HaileyStormC That is not correct. See Appendix A for olympiad benchs and o1-preview GPQA Diamond scores, also compared with hypothetical o1 not available for testing.
[2410.01792] When a language model is optimized for reasoning, does it still show embers of autoregression? An analysis of OpenAI o1
“ChatGPT’s new canvas interface is a game changer. Just used it to create a tesseract/hypercube visualizer with ThreeJS. Loving the unified UX — chat, inline comments, and watching GPT-4o work its magic on the code — all in one place. Does not get old. Claude artifacts have some
“Sam Altman says o1-preview is “deeply flawed,” but o1 will be better!! A few months ago, he said GPT-4 sucks and that the next model will take everyone’s breath away. 🙄 I guess since everything sucks or is deeply flawed, we shouldn’t be using any of his models, right? 🤣” / X
“🤔 I read this differently than the thread suggests: – o1-preview is clearly in a league apart: solves 2/6 variants, and on the rest gets around 50%; all other models got less than 10 – o1-mini is stem optimized and this is not stem, makes sense it fails. I say congrats o1 team” / X
“As usual, OpenAI failed to emphasize the real-game changer feature at their Dev Day: audio output from the standard generation API. This has severe implications for text-to-speech apps, particularly if the audio output style is as steerable as the gpt-4o voice demos.
“🚨 OpenAI just dropped a new open-source model 🚨 Whisper V3 Turbo is a new Whisper model with: – 8x faster relative speed vs Whisper Large – 4x faster than Medium – 2x faster than Small – 809M parameters – Full multilingual support – Minimal degradation in accuracy
Whisper Large V3 Turbo WebGPU – a Hugging Face Space by webml-community
“🚀 OpenAI’s new Whisper “turbo”: 8x faster, 40% VRAM efficient, minimal accuracy loss. 🔒 Run it locally in-browser for private transcriptions! Transcribe interviews, audio & video. ⚡️ 40 tokens/sec on my MacBook 🔗 Try it:
“o1 has sparked tons of ideas for applying LLMs to reasoning problems in science and math, but one of the most interesting applications IMO is prompt optimization… TL;DR: Prompt engineering is a black box even for recent frontier models–slight changes in prompts lead to big
“🎶 Realtime voice agent in TypeScript 🤖 We’ve added a TS example of a voice agent built using @OpenAI’s new Realtime API! It’s a @honojs 🔥 app with a default @TavilyAI websearch tool. Extend it with any prebuilt LangChain or custom tool. Repo:
Building agent + human collaboration with GPT-4o | OpenAI
Mira Murati’s Departure Is the Latest Drama at OpenAI – The New York Times
OpenAI Chairman: AI Is in a Bubble – Business Insider
OpenAI Training Data to Be Inspected in Sarah Silverman Copyright Case
The Art of the OpenAI Deal
OpenAI is Revamping Sora AI Video — The Information
“OpenAI canvas is interesting, because it moves away from pure chatbots to a new angle that feels less like ordering around in intern and more like working with a co-worker But I do wonder how much work will be done inside this interface, as it doesn’t fit into current workflows” / X
Introducing canvas, a new way to write and code with ChatGPT. | OpenAI
“For the first time we are fundamentally changing how humans can collaborate with ChatGPT since it launched two years ago. We’re introducing canvas, a new interface for working with ChatGPT on writing and coding projects that go beyond simple chat. Product and model features:
OpenAI launches new ‘Canvas’ ChatGPT interface tailored to writing and coding projects | TechCrunch
“check out canvas in chatgpt:
OpenAI’s ChatGPT Breaks Out of Its Box—and Onto a Canvas | WIRED
OpenAI Is Growing Fast and Burning Through Piles of Money – The New York Times
OpenAI’s DevDay brings Realtime API and other treats for AI app developers | TechCrunch
New Credit Facility Enhances Financial Flexibility | OpenAI
“@herbertong @thetylerhayes OpenAI closed a new $6.6B funding round, now valuing the company at $157B. This solidifies its position as the most well-funded AI startup in the world. Nice. (Chart from @chartrdaily)
SoftBank to Invest $500 Million in OpenAI — The Information
OpenAI DevDay 2024 | OpenAI
“How structured outputs work under the hood (via breakout at OpenAI DevDay) Guess why the first structured output request is slow, but the 2nd+ is fast? Engineering: * Unconstrained token decoding isn’t good. The model could pick any token. * Limiting which tokens can be
New funding to scale the benefits of AI | OpenAI
“OpenAI will be sponsoring Marijn Haverbeke, the creator and maintainer of the two open source libraries used in making ChatGPT canvas, ProseMirror and CodeMirror. Excited to support Marijn’s work as we continue to make canvas better for developers! 🖼️🧑💻
OpenAI Discusses Giving Sam Altman 7% Stake in For-Profit Transition – Bloomberg
OpenAI CTO Mira Murati Is Leaving the Company | WIRED
“OpenAI cloned the functionality inside @cursor_ai i.e. the apply model. Nice idea, poor execution – it doesn’t work very well. Often updates the entire doc, not the selection – and when it does update the selection it doesn’t match it correctly. Just use Cursor 😃
Introducing the Realtime API | OpenAI
OpenAI Reportedly Slated for $500 Million SoftBank Investment
OpenAI Completes Deal That Values Company at $157 Billion – The New York Times
OpenAI asks investors not to back rival start-ups such as Elon Musk’s xAI
Prompt Caching in the API | OpenAI
OpenAI’s Original Sin – by Alberto Romero
OpenAI to remove non-profit control and give Sam Altman equity, sources say
“Trying something new: I rigged up a liveblog system and I’m using it at OpenAI DevDay in San Francisco right now
Altman reportedly trying to sell Biden on a slew of AI DCs • The Register
Model Distillation in the API | OpenAI
Leave a Reply