Toronto-based AI company Cohere has launched Embed 4, a new multimodal search model capable of interpreting text, PDFs, graphs, and other visual formats across more than 100 languages. Designed for enterprise use, Embed 4 outperforms OpenAI’s top embedding model in multilingual tasks and excels in sectors like finance, healthcare, and manufacturing.
Cohere’s business-facing focus contrasts with consumer models, emphasizing real-world decision support over image generation. “Most businesses don’t need more pictures of cats,” said co-founder Nick Frosst. The model has been optimized for performance in languages such as Japanese, Arabic, and French.
Want to know more? Check out the source code on The Logic.