Cohere Inc., an artificial intelligence firm, is launching a new AI model on Thursday. The model outperforms competitors like OpenAI on specific “business-critical” tasks at a far lower cost, intensifying the competition among AI businesses for corporate clients, according to the company.
Cohere claims that its new model, named Command R+, will be the startup’s most potent technology yet, after releasing a smaller but less functional version last month. With its AI technology, Cohere caters mostly to commercial clients who can utilize it for activities like data analysis and copywriting.
In contrast to other prominent AI firms, Cohere chose not to establish a customer-facing enterprise. Chief Operating Officer Martin Kon declared, “We do not have and never will have a cash-burning consumer chatbot.”
Martin continued, “Our lane is focusing on enterprise, and helping get our models into production at a massive scale. And this model is a huge step forward in terms of capabilities as well as distribution.”
According to Cohere, their new technology advances a technique known as retrieval-augmented generation. The method increases a chatbot’s accuracy and enables the tool to offer citations for its answers, which it claims reduces “hallucinations,” or inaccuracies, according to the company.
The model can execute tasks in ten languages, and it can do more than just generate text—it can also automate some commercial activities, according to the company.
It added that the price will be lower than that of OpenAI’s GPT-4 Turbo model. The most sophisticated models from Anthropic and OpenAI outperform Cohere’s technology overall, but the company claims that Cohere excels in certain business tasks like question answering and citation-based summarizing.
In the upcoming weeks, Cohere’s models will be made available on Amazon Web Services and other platforms, after initially being housed on Microsoft Azure. Currently, the startup’s clientele includes Accenture Plc, Oracle Corp., and McKinsey & Co.
Toronto’s Cohere, established in 2019, specializes in customizing large language models for users. CEO Aidan Gomez, formerly at Google, co-authored the pivotal 2017 AI paper “Attention is All You Need.”
The company aims to raise over $500 million at a $5 billion valuation, with $22 million in annualized revenue as of March. Revenue surged by 60% in the first 10 weeks of 2024, largely due to enterprise customers’ interest in its new models.
Also Read: BVM enables users to create AI models on the Blockchain
