A week ago, OpenAI released a new series of reasoning models in ChatGPT. Those new models (called o1-preview and o1-mini) spend more time thinking about problems before solving them, and the results are impressive. Here’s an analysis by LMSYS Org that shows how both o1-preview and o1-mini rank among other models (with o1-preview ranking #1 overall) You can start using both o1-preview and o1-mini today if you subscribe to ChatGPT Plus (read more about the release of those models from ... Read More

Meta AI launched a capability where creators can build an AI extension of themselves (or create an AI character based on their interests). And people can chat with these AIs in Instagram, Messenger, and WhatsApp (Meta’s products). This is available in the US only for now. Click here to learn more. Claude’s Artifacts Can Help You Maximize Your Productivity Using Claude 3.5 Sonnet with a new Claude AI tool called Artifacts can help you get the most out of the model. Artifacts help ... Read More

Meta just released its most advanced open-source AI model called Llama 3.1 with some impressive capabilities, including a 405B parameter model. You can read the full announcement from Meta by clicking here, but here are two quick takeaways: 1) Unlike closed models like GPT-4o and Claude 3.5 Sonnet, Llama 3.1 is open-source, which means that you can see how the model works and can build customized AI applications on top of it - this is huge for businesses. 2) Previous versions of ... Read More

Many business leaders are wondering if we’re in an AI bubble that’s about to burst. A while ago, David Cahn (a partner at Sequoia) published a great article entitled “AI’s $600B Question.” He questions where all the revenue is coming from to balance all that spending. He also gives some counterarguments to the “building railroads” analogy (that spending money on GPUs will eventually bring in all the trains), and warns that there’s a long road ahead. I highly recommend you read the ... Read More

Anthropic recently launched their latest model, Claude 3.5 Sonnet which is apparently better and smarter than all the other models, including OpenAI’s latest ChatGPT model (GPT-4o) and Google’s Gemini 1.5 Pro model. It also costs less per million tokens than Anthropic’s most intelligent model, Claude 3 Opus, as shown in this chart. It also has a 200K token context window. Here’s some benchmark data about how it performs compared to other ... Read More

Color Health is collaborating with OpenAI to develop a GPT-4-powered copilot application that helps clinicians identify missing diagnostics and create personalized cancer treatment plans. The copilot has improved healthcare outcomes by enabling clinicians to find four times more missing labs and other diagnostic results, drastically reducing delays in treatment. The reassuring takeaway is that every output generated by the copilot is reviewed and, if necessary, modified by a ... Read More

Interested in writing a nonfiction book but don’t have a lot of time? I just launched a new cohort-based course called the 4-Week "Write Your Book on the Side" System It's about how to write a high-quality, nonfiction book (in less than 4 weeks using AI tools) The course starts on June 25, 2024 and includes 8 total sessions over 4 weeks. Click here to learn more about The 4-Week "Write Your Book on the Side" System ... Read More

OpenAI just announced that all ChatGPT free users can now use GPTs. Previously, only paid ChatGPT could access them. This is going to be huge for businesses that have published (or will publish) public GPTs. 100 million weekly active users will now have access to those businesses’ GPTs, and that’s no small number. GPTs are custom versions of ChatGPT that can help you perform specific tasks. Some companies are already seeing incredible efficiency gains when they use them internally, ... Read More

The big news this week is that Microsoft will soon be enabling Copilot with their PCs (and according to Mustafa Suleyman, the CEO of Microsoft AI, this is a completely new product category). Basically, Copilot will soon work alongside employees in real time, regardless of what they’re doing (talking, typing, watching, etc.), which will make using AI seamless in the workplace. More from Suleyman about the announcement here: Don’t Hold Back on AI Training A study by PwC showed that ... Read More

During a live event this week, OpenAI released a new flagship model called GPT-4o, which can reason across audio, vision, and text. The demos about GPT-4o were pretty impressive, especially because ChatGPT can now process things in real time, such as helping you with a math problem while you’re actively trying to solve it. You can view a replay of the session by clicking here Allie Miller gave a great summary of GPT-4o’s capabilities in this thread Google I/O Unveils 100 Things ... Read More