- AI Tangle
- Posts
- ☕️ A Rivalry Between Mistral and OpenAI
☕️ A Rivalry Between Mistral and OpenAI
![](https://media.beehiiv.com/cdn-cgi/image/fit=scale-down,format=auto,onerror=redirect,quality=80/uploads/asset/file/079dbcb3-e858-4614-abc4-329735f02f42/image.png?t=1732573969)
Mistral AI aims to make the week exciting right off the bat with the release of the GPT-4-rivaling Mistral Large model, with Micron working on mass-producing chips for training AI models in the background. Meanwhile, Google shows off more Gemini-infused apps at the Mobile World Congress in Barcelona, and Meta gears up to tackle AI abuse and disinformation. Join us at AI Tangle as we untangle this week's happenings.
THE BIG AI STORY
Mistral AI, a French-based AI startup that has garnered significant funding despite its youth, is making waves in the industry with its latest announcements, introducing Mistral Large, the company's flagship large language model aimed at competing with the likes of GPT-4 and Claude 2 from OpenAI and Anthropic respectively. Alongside Mistral Large, the company is launching Le Chat, its alternative to ChatGPT, and partnering with Microsoft to provide Mistral models to Azure customers.
Let's elaborate a bit
Mistral AI's business model mirrors OpenAI's, offering Mistral Large through a paid API with usage-based pricing. The company's pricing strategy puts it in direct competition with established models like GPT-4 Turbo, boasting a 20% lower cost while still promising very neck-and-neck performance, according to Mistral AI's benchmark reports. Additionally, the launch of Le Chat signals Mistral AI's expansion into conversational AI, with plans for both free and enterprise versions, further solidifying the company's presence in the market.
6 QUICK HITS
Recently, Micron Technology has commenced volume production of its HBM3E (High Bandwidth Memory 3E) solution, set to be used in Nvidia's H200 video cards. These memory chips will boast up to an impressive 1.2 TB/s of bandwidth and a chip memory of 36 GB, though Nvidia's video cards will only be using the currently tested 24 GB, with sampling for the 36 GB variant beginning in March. While the company can't claim the throne of bandwidth speed, which goes to Samsung's 1.225 TB/s, Micron offers a noticeable decrease in power consumption for its memory chips and a much higher quantity in production.
At the Mobile World Congress in Barcelona, Samsung unveiled the Galaxy Ring, a health-tracking wearable providing health metrics and a "vitality score" through the Samsung Health app, which the company is considering a subscription model for. The Galaxy ring includes features such as monitoring heart rate, respiratory rate, sleep tracking, and providing an "AI health coach," which aims to take a deeper dive into a person's health. Hon Pak, head of the digital health team at Samsung Electronics, added that the company's Bixby virtual assistant may also be added to the Galaxy Ring.
Meta is gearing up to tackle disinformation and misuse of generative AI ahead of the European Parliament elections in June, aiming to establish an Elections Operations Center, including some of the company's experts from data science, engineering, research, and other teams, to identify and deal with potential threats in real-time. Meta intends to strengthen its partnership with fact-checking organizations and join forces with other tech firms to combat deceptive AI content, aiming to protect the integrity of elections worldwide.
Despite facing loud backlash with Gemini's AI image generation not long ago, Google introduced Gemini-powered features for Messages, Android Auto, and accessibility tools at the recent Mobile World Congress. The Lookout app now generates AI captions, while Maps gains improved screen reader support. Productivity enhancements include handwritten notes in Docs and Spotify integration, the Fitbit app updates offer data integration from various sources, and Wear OS adds Google Wallet pass access and voice-guided transit directions.
DeepX, an AI chip startup hailing from South Korea, recently announced that it had passed a $90 million funding milestone. Getting ready for global expansion and innovation acceleration, DeepX aims to meet the ever-increasing demand for AI-optimized semiconductors. As the industry forecasts exponential growth, DeepX's vision extends beyond monetary gains to revolutionize AI applications' efficiency and sustainability, navigating challenges with innovation and strategic partnerships.
San Francisco-based Interview Kickstart, a startup launched in 2014 focused on helping tech professionals hone their skills, secured $10 million in its first funding round from Blume Ventures. The company offers mentorship from top tech company workers, with a recent AI vertical becoming popular. Co-founders Ryan Valles and Soham Mehta offer courses aimed at preparing engineers for advanced roles, with learners achieving substantial salary increases. With a focus on sustainable growth, Interview Kickstart plans to expand its AI offerings and diversify into product management and design.
4 AI TOOLS
Zenfetch - Zenfech is an AI-powered personal search engine and assistant. Save, search, and interact with web content effortlessly, whether it be articles, videos, or PDFs.
Gaia - Find the right people quickly with Sylph's Gaia AI copilot system, a conversational search engine designed to help consumers connect with professionals, deep dive, and take action in moments.
Pitchleague - Get your pitch decks turning heads with Pitchleague, a platform powered by AI that analyzes and scores pitch decks, from structure to design, submitted by startup founders.
Persuva - Persuva aims to take your marketing and sales campaigns to the next level, allowing teams to easily craft tailor-made ads that speak to your audience as you know them.
AI READ & WATCH
A Conversational AI Revolution (16-min read OR 23-min listen)
In this podcast episode of the MIT Technology Review, Elizabeth Tobey, head of marketing, digital & AI at NICE, talks about the ever-evolving landscape of customer experiences and the beacon that AI has become to guide businesses toward seamless interactions.
Here to Stay (8-min watch)
Join as Jamie Dimon, CEO of JPMorgan, takes a deep dive into artificial intelligence at the High Yield & Leveraged Finance conference, stating his enthusiasm for AI's longevity and claiming it's not just hype anymore.