Inception, a G42 company, has introduced Inception 70B, a 70 billion parameter large language model (LLM) tailored for Arabic-based NLP solutions. This model is part of a comprehensive suite of 20 models, across 8 sizes ranging from 590M to 70B parameters, trained on 1.6T tokens of Arabic, English, and code data. The suite includes models small enough to run on a laptop and large models for enterprise applications. Dr. Andrew Jackson, CEO of Inception, emphasizes that these models aim to preserve Arabic heritage and democratize AI access. Neha Sengupta, principal applied scientist, notes that models up to 30 billion parameters were trained from scratch, while larger models presented greater computational and environmental challenges.