Consistent with the company’s commitment to open-source AI, the Granite models are released under the permissive Apache 2.0 license, making them unique in the combination of performance, flexibility and autonomy they provide to enterprise clients and the community at large.

The Granite 3.0 release reaffirms IBM’s commitment to building transparency, safety, and trust in AI products.

The Granite 3.0 technical report and responsible use guide provide a description of the datasets used to train these models, details of the filtering, cleansing, and curation steps applied, along with comprehensive results of model performance across major academic and enterprise benchmarks.

Raising the bar: Granite 3.0 benchmarks 

The Granite 3.0 language models also demonstrate promising results on raw performance.

The Granite 3.0 models were trained on over 12 trillion tokens on data taken from 12 different natural languages and 116 different programming languages, using a novel two-stage training method, leveraging results from several thousand experiments designed to optimize data quality, data selection, and training parameters.

IBM is also announcing an updated release of its pre-trained Granite Time Series models, the first versions of which were released earlier this year.

These new models are trained on 3 times more data and deliver strong performance on major time series benchmarks.

Introducing Granite Guardian 3.0: ushering the next era of responsible AI  

IBM is also introducing a new family of Granite Guardian models that permit application developers to implement safety guardrails by checking user prompts and LLM responses for a variety of risks.

The Granite Guardian 3.0 8B and 2B models provide the most comprehensive set of risk and harm detection capabilities available in the market today.

In addition to harm dimensions such as social bias, hate, toxicity, profanity, violence, jailbreaking and more, these models also provide a range of unique RAG-specific checks such as groundedness, context relevance, and answer relevance.

Assistants to Agents: realizing the future for enterprise AI 

IBM is advancing enterprise AI through a spectrum of technologies – from models and assistants, to the tools needed to tune and deploy AI specifically for companies’ unique data and use-cases.

IBM is also paving the way for future AI agents that can self-direct, reflect, and perform complex tasks in dynamic business environments.

IBM continues to evolve its portfolio of AI assistant technologies – from watsonx Orchestrate to help companies build their own assistants via low-code tooling and automation, to a wide set of pre-built assistants for specific tasks and domains such as customer service, human resources, sales, and marketing.

Today IBM also unveiled the upcoming release of the next generation of watsonx Code Assistant, powered by Granite code models, to offer general-purpose coding assistance across languages like C, C++, Go, Java, and Python, with advanced application modernization capabilities for Enterprise Java Applications.

Expanded AI-powered delivery platform to supercharge IBM consultants with AI

IBM is also announcing a major expansion of its AI-powered delivery platform, IBM Consulting Advantage.

The multi-model platform contains AI agents, applications, and methods like repeatable frameworks that can empower 160,000 IBM consultants to deliver better and faster client value at a lower cost.

As part of the expansion, Granite 3.0 language models will become the default model in Consulting Advantage.

Another key part of the expansion is the introduction of IBM Consulting Advantage for Cloud Transformation and Management and IBM Consulting Advantage for Business Operations.

Each includes domain-specific AI agents, applications, and methods infused with IBM’s best practices so IBM consultants can help accelerate client cloud and AI transformations in tasks, like code modernization and quality engineering, or transform and execute operations across domains, like finance, HR and procurement.

The “Granite 3.0” models, an arm of IBM’s Granite family, provide high performance and flexibility while being made available as open-source software, contrasting the strategy of competitors like Microsoft that charge for access to their models.

Unlike some rivals, IBM provides a paid tool called Watsonx, which allows companies to run and manage customised models within their own data centres. 

Variants of the Granite 3.0 models have now been made commercially available on the Watsonx platform, with some also integrated into Nvidia’s software tools, enabling broader accessibility for enterprises.

A key feature of the Granite 3.0 release is the use of Nvidia’s H100 GPUs in the training process, which supports these models’ capability to perform a range of enterprise-specific tasks. According to IBM’s research director Dario Gil, this collaboration helps ensure that the models are optimised for efficiency and cost-effectiveness in business settings.

The Granite 3.0 models, which come in various sizes and functionalities, offer extensive flexibility, performing tasks like data retrieval, classification, and entity extraction. They are especially suited for businesses looking to tailor AI systems to their unique data requirements. IBM highlights the models’ efficiency, noting they can match or exceed the performance of larger models from competitors at a fraction of the cost, benefiting enterprises seeking budget-friendly solutions.

In addition to general-purpose models, IBM has introduced the Granite Guardian 3.0 models, which focus on ensuring safety and reliability in AI applications. These models provide safeguards against risks like bias, toxicity, and other harmful content, which can be critical for enterprises aiming to deploy AI responsibly. The Guardian models also perform checks on AI-generated outputs to ensure relevance and accuracy, further reinforcing their reliability in business operations.

The release of Granite 3.0 models reiterates IBM’s focus on transparency and responsible AI use. The models have been trained on a massive dataset, spanning multiple languages and programming environments, and are designed to be highly adaptable for enterprise needs. IBM’s technical reports on these models also outline the steps taken to ensure data quality, safety, and performance, further establishing trust with clients looking to integrate AI into their workflows.

Again, IBM is expanding its AI-powered tools and platforms, with the Granite models becoming the default in IBM Consulting’s delivery systems. This will enable IBM’s consultants to use AI more effectively in assisting clients with cloud transformations, code modernisation, and operations management. As part of this expansion, IBM aims to improve the speed and efficiency of AI deployment in various sectors, enhancing productivity and offering solutions that are customisable to specific business challenges.

The Granite 3.0 models are available for download under the permissive Apache 2.0 license, enabling developers to integrate them freely into their projects. IBM’s collaboration with partners such as AWS and Google Cloud will further extend the reach of these models, allowing more businesses to benefit from their capabilities.

IBM is aiming to deliver flexible, high-performance AI models that are both cost-effective and targeted toward enterprise needs.