European AI company developing state-of-the-art language models with a focus on efficiency.
Mistral AI
Flagship 123B-parameter model from Mistral AI, progressively upgraded through 2024 versions. Targeted to be a top-performing open model competing with Meta's Llama family on knowledge and reasoning tasks.
A 123B-parameter **multimodal** model from Mistral, combining language with vision ("Pix-"). A 12B smaller version was released under Apache 2.0. Pixtral extends Mistral's LLM capability to image understanding and description tasks.
A sparse Mixture-of-Experts model (8 experts) with a total of 46.7B parameters (12.9B active per token). Outperformed dense models like Llama 2 70B and GPT-3.5 on many benchmarks at significantly lower compute cost.
7.3B-parameter model trained from scratch with improved data and architecture enhancements. Open-source (Apache 2.0) with strong performance that rivaled larger 13B+ models on many benchmarks.
