TechCrunch’s Post

One of the big trends this year has been SMLs (small language models). They're efficient and cheap to run. The performance capabilities overall have also increased for models of small-medium size. If those gains move up to larger models, it means trillion plus parameter models get a lot more bang for its buck too - driving exponential gains. (not even counting CPU driven LLMs... that's another game changer I'll cover next) #openai #apple #ai #sml #llm https://tcrn.ch/3zP3Pmn

OpenAI has launched GPT-4o mini, a new small AI model that promises greater affordability and speed compared to its existing models. Available now for developers and consumers through the ChatGPT web and mobile apps, and for enterprise users next week, GPT-4o mini is optimized for cost-efficiency and high performance in reasoning tasks involving text and vision. It replaces GPT-3.5 Turbo as the smallest model in OpenAI's lineup, boasting superior benchmark scores and being over 60% cheaper to run. Additionally, OpenAI has introduced new tools for ChatGPT Enterprise, including the Enterprise Compliance API, enhancing data management and control for admins in regulated industries. #TechInnovation #AI #OpenAI #GPT4oMini #Developers #EnterpriseTools #AIModels #ArtificialIntelligence https://lnkd.in/gxXfmrSt

OpenAI plans to launch a store for GPTs, custom apps based on its text-generating AI models (e.g. GPT-4), sometime in the coming week.🙌 read below to find out more about it! #chatgpt #ai #tech #technology #developer #webdevelopment #aimodels

Multimodal is the new buzzword in the world of AI. You can interact via text, image and audio in the same conversation. Tested out on OpenAI mobile app with the new GPT4o. Experience was not seamless but definitely able to interpret. Next step would be to integrate video in the model. https://lnkd.in/gk5UBswf

Today, OpenAI unveiled its latest innovation: GPT-4o mini. This smaller, faster, and more affordable AI model is poised to make waves in the tech community. 🌟✨ 📅 Starting today, developers and consumers can access GPT-4o mini through the ChatGPT web and mobile app, with enterprise users gaining access next week. 🗓️ more details: https://lnkd.in/dWFzb2by Key Highlights: 🔸 Speed & Efficiency: Ideal for high-volume, simple tasks, offering a practical solution for developers. 🔸 Performance: Achieves impressive scores—82% on MMLU and 87% on MGSM—outperforming competitors like Gemini 1.5 Flash and Claude 3 Haiku. 🔸 Cost-Effective: Provides a more budget-friendly option without sacrificing quality. As small AI models continue to evolve, GPT-4o mini stands out as a significant advancement, promising enhanced performance and efficiency for various applications. 🌐💡 #OpenAI #GPT4oMini #AIInnovation #TechNews #AIRevolution #Developers #MachineLearning #ArtificialIntelligence #ChatGPT #AIModel #Efficiency #TechUpdate

App Store for AI, customized digital assistants, AI agents. How will organizations govern and manage these assets, authorization, access, privacy, endpoints, security, data use? I hear a next phase of AI Ethics. #aiethics #aigovernance #cybersecurity #equityai

Get Ready For GPTs I am curious to see the combined potential of crowd-sourcing and artificial intelligence. OpenAI is set to launch the GPT Store, a platform for developers to list custom applications based on its AI text-generating models like GPT-4. Announced at the DevDay conference last year, its launch was postponed due to a leadership change in the company. Details about potential revenue-sharing or payment structures for developers are still unclear, although OpenAI plans to reveal more information soon. This move marks a shift for OpenAI from being just an AI model provider to a platform, allowing broader access to AI app creation and possibly impacting businesses built around creating similar applications. #ai #innovation #chatgpt #gpts

Here are some reflections on OpenAI omni model released yesterday. They put a lot of personality into the default voice and went all Samantha from the movie Her on it. There are other voices but it's a fairly big shift in how they've been presenting AI in the past. It went hand in hand with a big boost to their free tier (browsing, code interpreter, images, GPT store) which removes a massive barrier for educators using AI and not wanting to depend on stuff students need to pay for. Omni is a great coder and topped the aider llm leaderboard, though it falls behind claude 3 opus on the refactoring benchmark. It beats claude 3 opus on speed and price significantly so it's my default model. The api access is a lot slower than the web interface. That's pretty normal, they tend to put high rate limits on api usage at the beginning of a rollout but I wouldn't be surprised if the speed in the web interface will be dialled back a bit over time. I had access to the interactive chat in my phone app yesterday, but not today. I've also noticed their voice to text api is slower than user so it looks like a bumpy roll out. Strategically OpenAI continues to be at the front in terms of capabilities with other providers playing catchup, while also lowering their prices and increasing free features. This is the (more important) post of the bump to the free tier : https://lnkd.in/gXr7QHDe And their omni launch post : https://lnkd.in/gPyU6nMB

Last night, I found myself drawn into the same wonder I felt watching early iPhone keynotes more than a decade ago - only his time, it was for OpenAI's DevDay, and yes it was at 2AM here. The launch of the GPT Store is a game changer, feeling very much like when the Apple App Store was first announced. I can't wait for my first GPT creation - an Instagram chat game called Ah Beng GPT, to be played by so many more people..... lah. He's also the heart of Replyr, because the feedback we got from players was reason Vishvesh and I started this company. On the business side, I'm most excited about the unveiling of GPT-4 Turbo with vision, and a 128k token context window. In simple English, it's AI that remembers 16x more information, can see things and can respond way faster: 3 enhancements we've been wracking our brains trying to figure out for our clients. We've started testing it, and yes it does live up to the hype vs. regular ol' GPT-4! Thanks for making our lives so much easier OpenAI! ps: anyone can try the new APIs for themselves, sign up on OpenAI and build a GPT-4 Turbo powered chatbot on BotDistrikt like we did.

OpenAI is launching the GPT Store next week! 🚀 What does this mean exactly...? 🤔 Now, anybody will be able to customize a ChatGPT bot to their own needs and publicly list it, without the need to write a single line of code - just using text commands! 🧑💻 Imagine the power of GPT-4 tailored to your company's data and needs. 🌐💡 #OpenAI #GPTStore #Innovation #AI #NoCode #ChatGPT #GPT4 #TechnologyTrends" https://lnkd.in/eUc7XVnG