IBM launches Granite 3.0 family of high-performing AI models for businesses

At its recently held TechXchange event, IBM announced the release of its most advanced family of AI models yet - Granite 3.0.

With the release of Granite 3.0, IBM reaffirmed its commitment to building responsible AI. (Express Image/IBM)

IBM has unveiled a set of high performing AI models built for businesses. The new Granite 3.0 8B and 2B models have been released under Apache 2.0 license (a permissive free software license) showing strong performance across numerous academic and enterprise benchmarks. The company claims that the models have been able to outperform or even match similar-sized models.

The new AI models were introduced at IBM’s annual TechXchange event. Claimed to be IBM’s most advanced family of AI modes to date, Granite 3.0 showcases strong performance, safety, and transparency. The AI models align with the company’s commitment to open-source AI.

The Granite 3.0 family includes General Purpose/Language models such as Granite 3.0 8B Instruct, Granite 3.0 2B Instruct, Granite 3.0 8B Base, and Granite 3.0 2B Base. When it comes to guardrails and safety, the family includes models like Granite Guardian 3.0 8B and Granite Guardian 3.0 2B. The other models in the family are Granite 3.0 3B-A800M Instruct, Granite 3.0 1B-A400M Instruct, Granite 3.0 3B-A800M Base, and Granite 3.0 1B-A400M Base.

Story continues below this ad

Also Read | ‘Trust and governance must be at the core of enterprise AI’: IBM’s Sriram Raghavan

Talking about the new family of models, Viswanath Ramaswamy, VP, Technology, IBM India & South Asia said, “As Indian businesses embark on their generative AI journey, many are limited by the data available to today’s LLMs, with most enterprise data remaining untapped. Our new Granite models address this by offering truly open and cost-efficient AI solutions, optimised for tasks like RAG, classification, summarization, and entity extraction.”

He also said that combined with InstructLab, Granite 8B models enable businesses to integrate enterprise data into LLMs at up to 23x cost savings in areas such as finance, HR, and customer support.

According to the company, the Granite 3.0 8B and 2B language models have been designed as ‘workhorse’ models for enterprise AI and they deliver strong performance for tasks such as classification, summarisation, entity extraction, Retrieval Augmented Generation (RAG), and too use. The company has said that the models are highly versatile and have been designed to be fine-tuned with enterprise data and can be seamlessly integrated across diverse business environments or workflows.

Although most Large Language Models are trained on publicly available data, a vast majority of enterprise data remains untapped. IBM with its slew of powerful AI models can help businesses achieve task-specific performance on part with larger models at a faction of the cost.