Loading…
AMD's Instella is a family of fully open, state-of-the-art 3-billion-parameter language models trained from scratch on AMD Instinct MI300X GPUs. The models aim to deliver high performance and outperform existing fully open models of similar size. The Instella-3B-Instruct variant provides instruction-tuned capabilities for conversational and text-generation tasks and is available on Hugging Face.
AI named Instella in December 2025.