By Krystal Hu
Canadian AI startup Cohere, last valued at $5.5 billion, will focus on creating tailored models for enterprise users rather than broader base models, the company told Reuters.
The evolution of its strategy, presented Thursday in a letter to its investors, comes as many companies are still trying to understand how to integrate large language models into their daily work, two years after the arrival of ChatGPT.
“What we’re hearing from customers is that they don’t just need bigger models to be good at everything. They need models that are actually designed for their specific use cases,” he said. said Nick Frosst, co-founder of Cohere, in a statement. interview with Reuters.
Cohere, seen as a competitor to AI labs including OpenAI and Anthropic, says it will continue to develop basic models but will focus on other training techniques to improve the models, instead of ‘increase the size of the models. Although selling application programming interfaces (APIs) to its models will remain a small part of Cohere’s offering, the focus is on deploying custom models.
The race to build bigger and better models has fueled an investment boom, from startups to big tech. OpenAI, Anthropic and xAI have raised billions to fund the capital-intensive development of cutting-edge AI models.
Headquartered in Toronto and San Francisco, Cohere has raised more than $900 million from investors including Nvidia, Cisco and Innovia Capital.
Cohere presented itself as an enterprise-focused AI company, independent of cloud providers. The company works directly with customers such as Oracle and Fujisu to tailor models to specific needs.
Cohere’s new focus also comes as the industry, which had seen breakthroughs in increasing computing power and model size, is seeing efficiencies from larger models decline. AI labs face delays in training the next generation of large language models. Ilya Sutskever, co-founder of AI labs Safe Superintelligence (SSI) and OpenAI, recently told Reuters that the results of the pre-training scale-up have stagnated.
Frosst said that simply increasing the size of the model doesn’t always yield better results. The focus on customization could allow Cohere to be more capital efficient, reducing the need for computing power. The company is not pursuing artificial general intelligence (AGI) like OpenAI.
“We’re going to work with a company to figure out how to make the model perfect for their use case, tailor it to their specific needs, and move into production, not counting on the future of AGI coming next year ” Frosst said. .
(Reporting by Krystal Hu in New York; editing by Elaine Hardcastle)