Introduction to Arcee Foundation Models

Arcee Foundation Models (AFM) are a family of instruction-tuned generative models purpose-built for enterprise-grade use. Unlike general-purpose large language models, AFMs are optimized from the ground up for:

• Cost efficiency at scale

• Customizability to domain-specific needs

• Compliance with enterprise data policies and legal frameworks

AFMs are open architecture, open weight models made available with Apache 2.0 licensing. Whether you're deploying in the cloud, on-premise, or on-device, AFM models are designed to run efficiently in diverse environments without compromising performance or compliance.

Key Features

• Built for Enterprises: AFMs deliver strong performance across real-world business tasks at a fraction of the cost of frontier LLMs.

• Deployment Flexibility: Run on cloud clusters, edge devices, or local CPUs/GPUs using Hugging Face, vLLM, llama.cpp, etc.

• Clean Data, Compliant Foundation: Trained on high-quality, rigorously filtered data.

• Multilingual & Modular: Support for major global languages with plug-in adaptation for more.

• Open Source Licensing: Apache 2.0 to give unmatched flexibility to startups, researchers, and enterprises.

Available AFMs

The first AFM released is AFM-4.5B. For details on the model and deployment quick starts, proceed to AFM-4.5B.