Arcee AI

We are honored to be recognized as a 2025 Intelligent Applications 40 winner! The #IA40 recognizes the top private AI and intelligent application companies shaping the future. Thank you, Madrona, Amazon Web Services (AWS), Microsoft, Google, NYSE, Delta Air Lines, Goldman Sachs, and McKinsey & Company for the recognition. See the full list of winners and request an invite for the #IASummit happening on Oct. 1: https://www.IA40.com -- Follow Arcee AI for the latest news on small language models.

Small language models, such as AFM-4.5B, and Arm-based CPUs are a great match. We recently published two extensive tutorials on the Arm website. You'll learn how to set up either an Amazon Web Services (AWS) Graviton or a Google Axion instance, download and optimize the model, run inference, and evaluate performance and perplexity. You’ll be surprised by the numbers! ➡️ AFM-4.5B on AWS Graviton: https://lnkd.in/e3drhvZu ➡️ AFM-4.5B on Google Axion: https://lnkd.in/eMUYqsjA ➡️ AFM-4.5B on Hugging Face: https://lnkd.in/eGXuSwAa https://lnkd.in/etbVCF6G https://lnkd.in/e3YGYm3C -- Follow Arcee AI for the latest news on small language models

We're happy to be working with Amazon Web Services (AWS) and we're looking forward to doing more together!

Our recent webinar with Intel Corporation is now available at https://lnkd.in/eJxayDcQ In this session, you’ll learn how Arcee Foundation Models help decision makers to: • Achieve state-of-the-art AI performance optimized for Intel Xeon CPUs • Reduce the total cost of ownership by avoiding expensive GPU requirements • Maintain data privacy, compliance, and sovereignty for regulated environments • Leverage an open, transparent foundation model for maximum control • Accelerate deployment with AFM’s domain adaptation for industry-specific solutions -- Follow Arcee AI for the latest news on small language models

Many AI models are like specialists: they excel in one area but are less effective outside of it. Others are great thinkers, able to reason step by step, but don’t know much about specific fields. What if we could have both in one model? However, when you merge a “thinking” model with a “specialist” model, one skill often gets lost. The result? Weaker performance or even nonsense answers. That’s the challenge a team at South China University of Technology tackled. Their solution, RCP-Merging, preserves the reasoning skills while incorporating deep knowledge. In tests, it boosted medical AI accuracy by 9.5% and finance AI accuracy by 9.2% — beating all other methods. To make it happen, they turned to Arcee AI MergeKit to quickly and fairly test various methods of model merging. MergeKit made it possible to demonstrate that RCP-Merging is effective across multiple AI architectures. The takeaway: With the right tools and method, we can have AI that both thinks deeply and knows its stuff. Research paper: https://lnkd.in/eupZGhuc -- Follow Arcee AI for the latest news on small language models.

Don't forget our session with Intel Corporation today at 9:00 AM PDT. You can register now at https://lnkd.in/eJxayDcQ We've also published a product brief on the Intel website at https://lnkd.in/dNThjeJQ -- Follow Arcee AI for the latest news on small language models.

Join Arcee AI and Intel Corporation on August 12, 9:00 AM PDT, to learn how Arcee Foundation Models deliver high‑performance AI on Intel Xeon: no GPUs required. We’ll share how enterprises can: - Reduce AI infrastructure costs - Ensure privacy, compliance, and sovereignty - Deploy domain‑adapted models at scale - Speakers: Davis Stone, Julien SIMON, Shannon Warner Register now at https://lnkd.in/eJxayDcQ #GenAI #EnterpriseAI #IntelAI #ArceeAI -- Follow Arcee AI for the latest news on small language models.

VentureBeat is talking about the launch of our first foundation model, AFM-4.5B. Thank you, Carl Franzen! ➡️ VentureBeat article: https://lnkd.in/eekTtWpr ➡️ AFM-4.5B blog post: https://lnkd.in/e-JJudeN -- Follow Arcee AI for the latest news on small language models.

It's a busy week for Arcee AI. Hot on the heels of the launch of our first foundation model, AFM-4.5B (https://lnkd.in/e-JJudeN), we're equally delighted to announce the closing of a new strategic round. We want to welcome our amazing new investors to the Arcee AI family: ⭐ Prosperity7 Ventures, the venture arm of aramco ⭐ M12, Microsoft's Venture Fund ⭐ Hitachi Ventures ⭐ John Chambers' JC2 Ventures ⭐ Samsung Next ⭐ Guidepoint's CEO, Albert Sebag Thank you very much for your trust and support! ➡️ Blog post: https://lnkd.in/eeQB_pty -- Follow Arcee AI for the latest news on small language models.

Today, we’re excited to officially take AFM-4.5B out of preview and release the weights of both AFM-4.5B and AFM-4.5B-Base on Hugging Face. This marks a major milestone for our team at Arcee AI as we open up access to a new, enterprise-grade language model designed for both flexibility and performance across a wide range of deployment environments. Blog post: https://lnkd.in/e-JJudeN