IBM announced on Tuesday (May 21) that it will release a series of artificial intelligence models in the form of open source software.
In addition, IBM also revealed its collaboration with the Saudi Data and Artificial Intelligence Authority (SDAIA) on Tuesday. IBM will use its Watson platform to train SDAIA's Arabic model "ALLaM", which will also add new language skills to IBM's products, including the ability to understand multiple Arabic dialects.
Open source models available for customization
One year after the launch of IBM's artificial intelligence and data platform Watson, the company announced on Tuesday a series of updates to the platform and unveiled the latest open-source model, the Granite series.
This company does not strictly protect its artificial intelligence models like its competitors such as Microsoft, and charges fees for using these models; Instead, a similar approach to Meta Platforms was adopted, which opened up access to the model itself.
The company's latest open Granite series of artificial intelligence models can be customized for enterprises. One of its major features is to help software developers complete computer code faster.
According to IBM, Granite's large language models range in size from 3 billion to 34 billion parameters and are capable of performing a range of tasks, including code generation, fixing errors, interpreting and logging code, modernizing applications, and maintaining repositories. It is worth mentioning that its core model has been trained on up to 116 programming languages, which is why Granite is able to achieve "state-of-the-art" performance in coding tasks compared to other large language models.
IBM CEO Arvind Krishna stated in an interview that open source will become a key element of IBM's generative artificial intelligence strategy.
He also pointed out, "We believe we are in the early stages of generative artificial intelligence models. Ultimately, competition is beneficial for buyers. We also want to take responsibility for security."
As for how to generate revenue from open-source models, IBM has stated that it provides a paid tool on the Watson platform, allowing customers to run customized models more smoothly within their data centers.