Cohere

We’re excited to introduce our latest state-of-the-art model: Command A, which combines maximum performance with minimal compute! Command A is optimized for enterprises that require fast, secure, and highly efficient AI. It’s on par or better than leading models across agentic tasks, with a fraction of the compute required. You can run it with just two GPUs, ensuring hardware access isn’t a barrier to adoption. It features:   - Across the board improvements with business, STEM, and coding tasks  - A context length of 256k for enhanced RAG with citations on longer documents  - Expanded multilingual coverage for 23 languages for global teams - Agentic tool use to automate complex tasks   - Private deployment options for regulated industries Our next generation of models will help power a range of applications through our secure AI agents platform, North. It’s available today on the Cohere platform, Hugging Face, and coming soon to major cloud providers. Learn more: https://lnkd.in/g2jDaaWj

Our co-founder Ivan Zhang visited DC this week to join leaders from our partners Oracle and Accenture for a panel on focused on how AI is transforming federal operations, with insights drawn from commercial applications. Ivan shared key perspectives on AI governance, and practical use cases for the public sector, breaking down how Cohere meets the stringent security and privacy requirements in government. We're committed to delivering practical and secure technology solutions tailored to the federal sector.

What are the advantages of building models and products? A better, more customizable experience for customers across the stack. CEO Aidan Gomez joins Redpoint's Unsupervised Learning Podcast for a wide-ranging conversation on model architectures, enterprise adoption, and the benefits of vertical integration. https://lnkd.in/gYX9wBvi

We’re on the ground at the World Summit AI alongside our incredible partner, Fujitsu! Our collaboration with Fujitsu has led to groundbreaking innovations like Takane, the industry-leading Japanese LLM which sets the standard for accuracy and reliability. But we didn’t stop there. Our partnership has expanded to deliver global digital transformation solutions, combining Cohere’s state-of-the-art AI technology with Fujitsu’s robust solutions and services engine. Together, we’re focused on providing highly secure AI solutions to ensure our customers’ data is always protected.

Excited to announce that Cohere and Cohere Labs models are the first supported inference provider on Hugging Face Hub! ✨ Looking forward to this new avenue for sharing and serving our models, including the Aya family and Command suite of models. We're grateful - as always! - to collaborate with our Hugging Face friends. Thanks especially to Julien Chaumond, Simon B., Loucain Pouget, Célina Hanouti, Ben Burtenshaw, Merve Noyan and Vaibhav Srivastav. Read the announcement here: https://lnkd.in/g45RNKx2

🎉 Cohere’s latest Command A and Embed v4 are now available in the Azure AI Foundry model catalog. From intelligent document Q&A to enterprise copilots and scalable search apps, these models are production-ready and developer-friendly. Explore now: https://msft.it/6047q7F1D

Cohere is among the very first to get access to CoreWeave’s NVIDIA GB200 NVL72 systems! We’re excited to be at the cutting-edge with some of the newest hardware, unlocking both model training improvements and the ability to run North, our AI agents platform, on a single Blackwell GPU, so our enterprise customers can start using this hardware for secure AI applications in their own environments.  Learn more about how we’re using some of the first Grace Blackwell superchips: https://lnkd.in/gxa7kx5E

We’re excited to introduce Embed 4: the optimal search engine for secure enterprise AI agents! Our latest state-of-the-art model enables businesses to add powerful search and retrieval capabilities to their AI applications – accurately and quickly pulling insights from complex, unstructured, and lengthy mixed-modality data sources. Embed 4 offers: - Best-in-class multimodal search for text and image data across PDF reports and presentation slides like tables, graphs, code, diagrams, and more - A breakthrough context length of 128K tokens to generate embeddings for documents 200 pages in length, like annual financial reports, detailed contracts, or product manuals - Leading multilingual capabilities in 100+ languages - Enhancements for security-minded industries like finance, healthcare, and manufacturing with domain-specific understanding and secure deployment options Embed 4 provides enterprises an unmatched foundation for their search and retrieval pipelines and RAG applications. Try it out today on our platform, Microsoft Azure AI Foundry, or Amazon Web Services (AWS) Sagemaker. Learn more in our blog: cohere.com/blog/embed-4 

How Cohere's embedding & re-rank models transform semantic search capabilities 🧠🔎⚡https://go.aws/429e1lt Saurabh Baji, CTO at Cohere, shares how their powerful AI models in #AmazonBedrock enable intelligent data retrieval & ranking for enterprise applications. #AWS #generativeAI

We're excited to be named on the Forbes AI 50 list again! This recognition is a testament to our team’s hard work helping enterprises unlock real value with secure AI solutions.