OctoAI

Great to see an independent evaluation by Alex Volkov featuring OctoAI's fresh offering of Llama3.1 70B. It shows OctoAI's performance leadership using mainstream cloud GPUs. It is all about striking the right balance of speed, quality and cost to deliver value to our customers. Behind the scenes is our custom LLM serving infrastructure that builds on years of work in AI systems from ML compilers up to cloud engineering. Customers can run Llama3.1 in our managed SaaS as well as our OctoStack private offering. WanB dashboard: https://lnkd.in/gf9g2b7x

Pass 16x MORE information into the new Llama 3.1 models! Now with context windows up to 128k, unlocking new possibilities for: Retrieval Augmented Generation (RAG) Document summarization Business intelligence reporting tools And more! Take your enterprise applications to the next level 👆 Try the Llama 3.1 herd on OctoAI: https://octoai.cloud/text Read more here: https://lnkd.in/gmWjKJHR #OctoAI #RAG #DocumentSummarization #BusinessIntelligence

Catch Jared Roesch next week for Statsig's Founders by Founders! For #SeattleTechWeek hosted by Madrona 🙌 Register here: https://lu.ma/pvczoic7

A birthday message from Luis Ceze! 🎂

🎉 A year older, a year wiser, and a whole lot more AI-savvy! We're celebrating our birthday and reflecting on the amazing journey that's brought us to where we are today. 🎁 🎉 Just in the past year, we've achieved some incredible milestones: ✨ Launched the OctoAI platform and held a launch party at AWS ✨ Introduced our Text Gen and Media Gen solution and welcomed amazing customers like HyperWrite (OthersideAI) ✨ Changed our name to OctoAI! ✨ Launched OctoStack and onboarded a second cloud (GCP) in April ✨ Connected with some of the top AI engineers and developers through hackathons, in-office events, and builder's roundtables ✨ Grown our team, celebrated milestones, and helped customers build products that bring joy to people's lives And this week got even better when, with the integration of Llama 3.1 models from Meta! Now available on octo.ai (make sure to give it a try!) Our mission to make AI more sustainable and accessible is possible because of the incredible people who've joined us on this journey. Thank you to our team, customers, and partners! ❤️

Today is OctoAI’s birthday! Its amazing how much we accomplished in a short few years. We pioneered ML compliers and now 100s of commercial and R&D efforts are built on those technologies. We built Octomizer, the first SaaS tool for model optimization and deployment in edge and cloud. And now we have a world-class customizable, efficient and reliable GenAI serving stack. But what makes me most proud is our team culture, which is as much about celebrating wins as it is working through the challenges. I love this company and this team ❤️. Happy Birthday, OctoAI! With a huge thank you to our investors and customers for supporting us and trusting us with their business! 🙏 🐙 🚀

This time last year there were 16,000 open source LLMs available on Hugging Face — and that seemed like a lot at the time! Yet people there were questioning when OSS models would be “really ready for primetime.” Since then the state-of-the-art has evolved at warp speed, thanks in large part to AI at Meta and their Llama family of models. Today OctoAI has hundreds of happy customers running Llama 3 in production and saving up to 60X vs. closed models. Perhaps most exciting are the encouraging quality benchmarks that reveal Llama 3.1 405B to be the most capable open-source LLM ever — competing with best-in-class closed models for math, function calling, and reasoning. Our customers are already testing it out, and you can too https://lnkd.in/dDaWVfJw

I am thrilled to share that the entire Llama 3.1 herd — 8B, 70B and 405B — is available on OctoAI today. Thank you AI at Meta for your incredible partnership and to all of our Octonauts for working fast to make these model available to our customers immediately. Llama 3.1 is full of high-value features we know they’re going to love — a massive 128K context window, native tool calling, and support for 8 global languages, including my native tongue Portuguese! For enterprise users, Llama 3.1 405B unlocks new possibilities: an open source model that delivers transparency and model control that doesn’t sacrifice on quality. We’re also really excited to offer the entire 3.1 herd, including 405B, on OctoStack private deployment for privacy and compliance sensitive use cases. Get started today for free at https://octoai.cloud.

Huge achievement by the AI at Meta team on launching the Llama 3.1 models!  The quality benchmarks look incredible, our customers are going to be really excited for the whole Llama 3.1 herd. Learn more and try them on OctoAI here: https://lnkd.in/giU-mP8S. Thank you AI at Meta for the models and the OctoAI team for the incredible work!

OctoAI is proud to introduce the full Llama 3.1 herd to our customers, featuring 8B, 70B, and 405B models! These cutting-edge models unlock new possibilities for AI natives and enterprises, with: ✍ Massive 128k context window 🌍 Support for 8 global languages 📞 Native tool calling The Llama 3.1 herd boasts the largest, most capable open-source LLM available, Llama 3.1 405B, which sets a new standard for quality benchmarks. Say goodbye to the tradeoff between privacy and quality! With OctoStack, enterprises can self-host a powerful LLM that meets or exceeds GPT-4 quality, without compromising security and compliance. Ready to experience the power of Llama 3.1? Get started with the entire llama 3.1 herd for free at octoai.cloud and contact us to request a no-cost POC for Llama 405B on OctoStack! #OctoAI #meta AI at Meta https://lnkd.in/gmWjKJHR