Fujitsu and Cohere have announced a new strategic partnership to provide Japanese-language enterprise AI services. Cohere has developed multiple enterprise models that are highly secure and tailored to the real world of enterprises, “delivering real-world impact to customers,” the company said.
Cohere's Command R+ model will be the basis for the planned model as it has features such as verifiable accuracy, multi-language support and automation tools. The private cloud deployment will be available as part of the agreement to help serve highly regulated organizations, including financial institutions, the public sector and the R&D sector.
Fujitsu will use Cohere's Embed and Rerank models, in addition to the Command R+ model, to create enterprise search applications and search augmentation generation (RAG) systems.
Commenting on the agreement, Mr. Coheer said:
“By combining our cutting-edge AI technology with Fujitsu's expertise and fine-tuning capabilities, businesses will gain a best-in-class LLM with advanced Japanese language capabilities, helping to improve productivity and efficiency.”
“We are pleased to collaborate with Fujitsu, a global leader in technology that drives digital transformation with effective business solutions. We look forward to further developing this strategic partnership and providing cutting-edge AI technology to businesses around the world.”
Command R+ was announced in April, and according to Cohere, the model performs roughly on par with GPT4-turbo on multi-language, RAG, and tool usage benchmarks, and is also much less expensive to run on Azure than GPT4-turbo.
Command R+ features a 128k token context window for longer inputs, advanced search expansion generation (RAG) with citations to reduce hallucinations, multilingual support in 10 major languages, and tooling support to automate advanced business processes.
If this is your first time hearing about Cohere and its model, you can try it out on the Cohere dashboard. It's free to use – just log in with one of your online accounts, such as Google, and get started.
