Toronto-based artificial intelligence (AI) startup Cohere, an enterprise-focused competitor of ChatGPT creator OpenAI, has unveiled its newest large language model (LLM), Command R+.
Cohere, which is reportedly in talks to raise a sizeable funding round at a $5-billion valuation, said the new LLM is built off of its recently debuted Command R, while further improving its performance to make it competitive with more expensive LLMs on “key business-critical capabilities.” Microsoft concurrently announced that Command R+ will be integrated into its Azure AI model catalogue through a new collaboration with Cohere.
Cohere CEO Aidan Gomez said in a statement that collaborations such as the one between Cohere and Microsoft can shape the future of enterprise-grade AI, and will amplify Cohere’s ability to scale its AI capabilities to enterprises globally.
Command R+ will be available to developers and businesses immediately and will become available on Oracle Cloud Infrastructure and other cloud platforms in the coming weeks.
Cohere says Command R+ improves response accuracy and provides in-line citations to help mitigate hallucinations, which is when a LLM mistakenly generates nonsensical or inaccurate outputs.
Command R+ also adds support for multi-step tool use, which Cohere claims will allow the model to combine multiple tools over multiple steps to accomplish difficult tasks and even correct itself if it fails to accomplish a task with a tool.
Cohere introduced tool use with the basic Command R, and allows developers to use the LLM to automate internal infrastructure like databases and software tools, or external tools like customer relationship managers, search engines, and more.
Cohere also touts Command R+’s multilingual uses, claiming the new model contains a tokenizer that compresses French, Spanish, Italian, German, Portuguese, Japanese, Korean, Arabic, and Chinese, making it less data intensive than tokenizers from Mistal and OpenAI.
“Multilinguality is something that is crucial for equitable utility of this technology,” Gomez said in a post on X. “We want our models to work for as many people, organizations, and markets as possible. We perform strongly across 10 languages and we’re eager to expand this further.”
Founded in 2019 by ex-Google researchers, Cohere has been raising large amounts of funding and is one of Canada’s startup unicorns. Last June, the company raised a $270 million USD Series C round a mere 16 months after raising $125 million in Series B funding.
The launch of Cohere’s new LLM comes shortly after The Information reported that a Canadian pension fund, PSP Investments, is going to co-lead the company’s next funding round with an expected raise of at least $500 million at a valuation of approximately $5 billion.
Last week at CIX Summit 2024, Gomez spoke on the difficulties of fundraising while avoiding investments from large cloud players, saying “it lets us preserve our independence, we don’t get locked into one ecosystem.”
Feature image courtesy Cohere.