IBM Unveils Granite 3.0 as 'Workhorse' AI Model

IBM has unveiled what it claims is the most advanced model of its Granite AI product line, which will help it outperform similar models based on the leading benchmarks in the AI large language model (LLM) market.

The vendor unveiled the Granite 3.0 8B[illion] and 2B language models on Monday. Based on current benchmarks used by academics and corporate creators, IBM claims the models can match or surpass the performance of similar models made by Google or Anthropic. The models intend to help support a range of enterprise use cases, including IT automation, business process outsourcing (BPO), application development and cybersecurity.

According to IBM, this model's strength compared to others is that it was trained on 12 trillion tokens of data. Tokens are the smallest units of data in an AI model, and IBM's are drawn from 12 spoken languages and more than 116 coding languages.

IBM's Rob Thomas

“Part of our advantage in building models is data sets that we have that are unique to IBM,” SVP and chief commercial officer Rob Thomas said. “We have a unique ... 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.”

Another key differentiator is how IBM has releasd Granite 3.0's LLM under the Open Source Initiative (OSI)-approved Apache 2.0 open-source license. This will allow IBM partners to build their own brand and intellectual property on top of IBM's Granite models, and will empower a robust ecosystem of applications and solutions built on the LLM.

Related:Vultr, AMD, Broadcom, Juniper Team for ‘AI Innovation’

“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 company emphasized the partners' role in ensuring others had access to IBM's product line.

IBM's Kate Woolley

"The IBM Ecosystem acts as the linchpin to ensure clients can access and put AI advances like these to work for their business," Kate Woolley, GM of IBM's Ecosystem, wrote in a blog. "Comprised of [independent software vendors], hyperscalers, resellers and distributors, service providers, technology partners, consultancies, systems integrators and MSPs, these partners are bringing our state-of-the-art technology to business users at scale."

Granite is now available on several websites and platforms for partners to incorporate as they see fit, Woolley said. These include the AWS Marketplace, Qualcomm's AI Hub, Samsung SDS and parts of Nvidia's stacks.

Related:CF20: 2024's 20 Top UCaaS, CCaaS Providers Making the Most of AI

Channel Futures TV: The Coffee with Craig and James podcast talks with IBM's Kate Woolley about AI and the company's new MSP partner program in a recent episode.

Granite 3.0's Applications in 2024

IBM also released the next generation of the Watsonx Code Assistant, which offers assistance for general-purpose coding and new tools for building and deploying AI applications. Granite code models power the assistant and can help with coding languages like C, C++, Go, Java, and Python.

The IBM Consulting Advantage delivery platform was also updated with Granite 3.0. The new model's AI-powered delivery platform empowers more than 160,000 IBM consultants to deliver better and faster client value via assistants, the company said.