IBM Advances Open Source AI Strategy with New Granite Models and AI Assistants

Tech Giant Expands Watsonx Platform to Drive Enterprise AI Innovation

IBM has taken a significant step forward in its AI strategy by open-sourcing its Granite large language models (LLMs) and enhancing support for third-party AI models. These announcements came ahead of the IBM Think 2024 conference, reinforcing the company’s long-standing commitment to open source, which began in 2000.

Amidst high-profile releases of multimodal generative AI chatbots from OpenAI and Google, IBM's latest offerings may not be multimodal, but they provide access to third-party AI models that boast these capabilities. IBM positions itself as a neutral advisor for businesses embarking on their LLM journey, offering a platform designed to be adaptable to the rapidly evolving AI landscape.

IBM introduced two new AI Assistants: Watsonx Code Assistant for Enterprise Java Applications, which will be generally available in June, and Watsonx Code Assistant for Z, set to launch in October. These tools aim to accelerate the modernization of mainframe applications and provide natural language explanations for code.

The Granite LLMs, now available under the open-source Apache 2.0 licenses on Hugging Face and GitHub, range from 3 billion to 34 billion parameters. These models are designed for tasks such as large-scale application modernization, code generation, bug fixing, and code documentation. Trained on 116 programming languages, the Granite models are among the top-performing open LLMs for coding tasks.

With these advancements, IBM continues to solidify its role as a leader in AI innovation, supporting businesses in leveraging the power of AI to drive transformation and growth. The Granite models and AI assistants mark another milestone in IBM’s mission to make cutting-edge AI accessible and practical for enterprises worldwide.