Fast Company的动态

Really rough math: I estimate OpenAI spent $4,248,328,767 to run #chatgpt in 2023... that's $4.2 Billion with a capital B! On roughly $1.6B in ARR as of December... that is a staggering loss of $2,600,000,000 ($2.6B) I couldn't find actual spend numbers for 2023 but here is my math: 2022 Loss of $540M with $28M in Revenue = total spend of $568M That is $1,556,164 per day (conservative as it is the average for the year and the run rate exiting the year was probably much higher) Assume usage and cost 2X every 90 days (conservative?) So Q1: $283,221,918 Q2: $566,443,836 Q3: $1,132,887,671 Q4: $2,265,775,342 FY: $4,248,328,767 I am not saying that is good or bad I am just saying that OpenAI's transaction and dollar volumes are insane! OpenAI is now valued at about $100,000,000,000 ($100B) making it the most valuable US start up behind SpaceX and has raised $13B from Microsoft. https://lnkd.in/g6HN7b_v #openai #chatgpt #ai #futuretech

Explore the game-changing features of OpenAI's GPT-4o and discover how RAIA's RAMP Platform eases the way businesses manage these advancements.

#Topics OpenAI Aims for a Bolder Presence in the Developer Arena [ad_1] OpenAI, a major player in the artificial intelligence sphere, is gearing up to introduce significant updates aimed at enhancing the developer experience. Sources from Reuters have hinted that these changes will not only simplify the development process but also make it more cost-effective for developers to craft and market AI-driven applications using OpenAI's suite of technologies.Stateful API: A Game-Changer in Cost EfficiencyAmong the anticipated innovations is the Stateful API, a feature designed to retain request histories using added memory. This implies that documents input into the system wouldn't need to undergo expensive reprocessing with every new interaction. Insiders suggest that such an advancement could potentially reduce the operational costs of certain applications by a factor of up to 20. However, the exact savings are expected to vary depending on the application's nature.Introducing the Vision APIOpenAI is not just stopping at chat interactions. Another significant rollout on the horizon is the Vision API. This tool will empower developers to conceive scalable solutions geared towards image analysis and elucidation. The introduction of the Vision API is seen as a stepping stone to...

The new model from OpenAI not only enhances performance but also significantly lowers costs. How are businesses leveraging these advancements for strategic advantage?

This week, I watched the recent 20VC interview with Sam Altman & Brad Lightcap of OpenAI. For people curious about Ai, there exists no better way to learn about the industry than to watch video of leadership, watch their non-verbals, think about the implications of their words, dissect their business goals. Some interesting bullet points that I took away from this interview; - Enterprise adoption of ChatGPT/OpenAI tools will be faster than expected - There exists a mindset that AI/LLM models will or will not get better, many startups seem to be in the “will not get better” camp - The wrong way to think about GPT usage in enterprise space is to focus on a specific business metric and say “we will only use GPT if we believe we can increase or decrease this metric” - At inception, OpenAI leadership kept working on their stack because 1. it seemed like if they were successful it would be massively impactful and 2. they continued to make encouraging progress - They thought they would originally use AI tools for robotics and white collar task work and the current dominant use area tends to be more creative. - Sam Altman projects the vision that he is working towards a future that is “A world of abundance” My personal take on some of these bullet points, GPTs are deeply creative and flaws are a part of creativity, encourage your teams to embrace creativity and the flaws, using AI tools will shift the nature of work from doing, to sensing, observing, organizing, sequencing, arranging and calibrating, everyone's work will shift away from work to arranging work to be done. As related to “Ai models will or will not get better” ask yourself, did you predict that ChatGPT would come to exist or would be this useful/interesting? Pretty much no one did, not even Sam Altman, assuming the models won’t get better just seems weird given how little most people tend to know about what is possible and the quality of talent working on these problems. As related to “a world of abundance” I’m deeply skeptical of the counter-signaling that comes from San Francisco and I’m not even sure that abundant resource would be good for humanity, I do think that the focus on “as long as we create value in excess of cost AI will be worth it” is a good north star. Will OpenAI maintain dominance of the GPTs, generally I think is not useful to think about, competition among the players will benefit everyone, find the things that matter to your mission and form some deep feedback loops, the rest doesn't really matter. 20VC is an incredible platform for learning about the thinking, behavior and strategies of top global operators. I would rate myself as highly supportive of using OpenAi tooling for all manner of creative and product engineering work. https://lnkd.in/g3aD4rVj

Zuck is so vastly ahead of OpenAI, Anthropic… OG founder / CEO still running his ship with unilateral dual stock authority. ? Not outsourcing his data centers to anyone.. vertically integrate. ? All-in to spend tens of billions building the leading models and open weights them for everyone. No need to monetize models directly. ? Already created industry-standard LLM suite that even the top Transformer author (Noam) just defaulted to and gave up on building closed models. ? Endowed with a $50B + / year profit-printing machine to out-invest and out-build any upstart (OpenAI will lose > $10B over the next year). ? Can instantly deploy Llama (closed and open versions) across several BILLION active users continuously. ? ~ Joseph Jack (via the ?? platform) Zuck’s approach definitely shows a strong commitment to control and innovation. He transformed from the most hated CEO in Silicon Valley to one of the most interesting. Unfortunately, Elon took the opposite path. With stuff like this happening, and then OpenAI imploding, nothing makes sense. If OpenAI is being outpaced, what’s the deal with all of the fighting? It seems like maybe they found something big, and now everybody is fighting each other. I’m not sure I believe that OpenAI is behind. I don’t know if being behind would cause them to fight like this with different cofounders leaving. It doesn’t make sense to me! (via the ?? platform)

GPT-4 Turbo is now GA (Generally Available) As organizations worldwide continue to embrace the potential of AI-driven solutions, GPT-4 Turbo (the technology that powers OpenAI) stands at the forefront, offering unprecedented capabilities and opportunities for innovation. Whether it's crafting compelling marketing copy, personalizing customer interactions, or advancing scientific research, the possibilities with GPT-4 Turbo are limitless. Key features include: ?? Increased Language Understanding and Response ?? Enhanced Creativity in generating content that is engaging and imaginative ?? Improved Efficiency with optimised algorithms and compute enabling faster response ?? Customisation options that enable businesses and developers to seamlessly it integrate into existing workflows. Now GA means you're backed with SLAs. So what are you waiting for?

I’ve been banging on about OpenAI's (lack of) defensibility for a while. However I am now convinced by the latest release of Mistral AI's models that its latest $100B valuation is going to be a huge bonfire of investor cash. For those who haven’t been following all the model providers, Mistral is a tiny company - 37 employees on LinkedIn! - that raised Europe’s biggest ever seed round of €105m. This provoked a lot of derision (which I basically agreed with at the time). Lots of “How on earth are quaint little Europeans going to challenge the US juggernaut”, etc. Mistral’s first model was a tiny but cool model with 7B parameters. It is a lot of fun to play with, but seemed a bit niche - basically a specialist model for running on your phone or in restricted environments. It’s the latest models that have taken my breath away. Mistral-small aka mixtral delivers GPT-3.5 level performance at faster speed in an open source model. Whatever people may tell you about Llama, this isn’t something I’ve personally seen before. And most impressively, mistral-medium, for now not open-sourced but available via the new Mistral API platform, delivers comparable performance to GPT-4, at faster speeds. Oh, and did I mention it’s less than half the cost of GPT-4? It’s early days and I haven’t had the opportunity to run a real benchmark, but I think this is terrible for OpenAI. They still have the best model, but their lead is less than a year and narrowing. The polish of their end-user products is much improved, but it’s hard to see that #chatgpt as an end user product justifies a 100x revenue multiple. That sort of thing relies on OpenAI's models being the backbone of every AI product made in the future. I'm now sure that won't be the case. Here at Harriet we love OpenAI and respect the awesome job they have done with popularizing LLMs. But the technical ease of switching model provider (because the current generation of LLMs are all stateless) and the constant hunger for faster, better and cheaper means we have ~zero loyalty. We’re already looking at how we can use Mistral to support use cases where both low latency and high quality are important.?I would be stunned if others feel a greater level of brand loyalty. I heard Mistral founder Arthur Mensch on the A16Z podcast the other day (shoutout to Steph Smith who even if she feels is shouting into a void, is getting through to at least some people) and it’s super clear that he understands the real nuts and bolts of what developers want from LLMs in a way that’s focused on the practicalities of building high-quality software. Honestly it's so refreshing after non-stop pontification about the future of humanity and boardroom shenanigans. Has anyone else managed to have a play yet with the Mistral releases? I feel genuinely stunned by the quality of what I’ve seen but I’d love someone to bring me down to earth ;) You can sign up here: https://lnkd.in/dyxpV5pk Something for a Christmas project perhaps?!

Did you know that #LLMs can be lazy? A new version of OpenAI's #gpt4 is out, with fixes to address this problem. And if you're working with #embeddings, you might be interested in OpenAI's updated embeddings models. Find out about these and more in my InfoQ news: https://lnkd.in/esfD353r

#Topics What’s next for OpenAI | MIT Technology Review [ad_1] Of course, that was what he said in September. With top talent now jumping ship, OpenAI’s future is far less certain than it was.? What next for Microsoft?? The tech giant, and its CEO Satya Nadella, seem to have emerged from the crisis as the winners. With Altman, Brockman, and likely many more top people from OpenAI joining its ranks—or even the majority of the company, if today’s open letter from 500 OpenAI employees is to be believed—Microsoft has managed to concentrate its power in AI further. The company has the most to gain from embedding generative AI into its less sexy but very profitable productivity and developer tools.? The big question remains how necessary Microsoft will deem its expensive partnership with OpenAI to create cutting-edge tech in the first place. In a?post on X?announcing how “extremely excited” he was to have hired Altman and Brockman, Nadella said his company remains “committed” to OpenAI and its product road map.? But let’s be real.?In?an exclusive interview with MIT Technology Review, Nadella called the two companies “codependent.” “They depend on us to build the best systems; we depend on them to build the best models, and we go to market together,” Nadella told our edit...