IBM is rolling out new AI models, which it claims outperforms other popular large language models (LLMs).
IBM has released the latest version of its artificial intelligence (AI) models for businesses that will help the technology giant in the race towards AI agents.
The company on Monday announced its Granite 3.0 AI models, which are open source and tailored for enterprise use cases such as customer service, IT automation, Business Process Outsourcing (BPO), application development and cybersecurity.
The company claims it outperforms AI models from Meta, Anthropic and Mistral AI and is tougher to jailbreak.
The Granite models were trained on 12 trillion tokens of data, including across multiple languages as well as code data. The difference with IBM’s large language model (LLM) compared to others is that it trains its data on enterprise data, rather than public data pulled from the Internet.
“We have a unique, I’d say, vantage point in the industry, where we become the first customer for everything that we build that also gives us an advantage in terms of how we construct the models,” Rob Thomas, senior vice president and chief commercial officer at IBM, said during a briefing with press and analysts.
‘The most open source’
He also said that the LLM family “is the most close to open source that anybody has released”.
The difference between IBM’s model and Meta’s is that IBM released the models under the Open Source Initiative (OSI) approved Apache 2.0 open-source license.
“We decided that we’re going to be absolutely squeaky clean on that, and decided to do an Apache 2 license so that we give maximum flexibility to our enterprise partners to do what they need to do with the technology,” said Darío Gil, IBM’s senior vice president and director of research, at a press conference.
The open source model also allows IBM’s partners to build intellectual property on top of the Granite models.
“It’s completely changing the notion of how quickly businesses can adopt AI when you have a permissive license that enables contribution, enables community and ultimately, enables wide distribution,” Thomas said.
The future for AI agents
The company also mapped out its ambitions for AI agents, which is software that can perceive its environment and is based on data that take autonomous actions.
“We’re at the very beginning of this era around agents and ideally assistants and agents is almost two sides of the same coin,” said Thomas.
One of the ways IBM said it would focus on AI agents was through domain-specific agents, which are designed to understand unique jargon and content for certain industries.
Human resources (HR) is one of its use cases and it has HR assistants that it is showcasing, said Ritika Gunnar, IBM’s general manager for data and AI.