Tech know-how to go: Open source vs. proprietary models of generative AI

Blog post -

Tech know-how to go: Open source vs. proprietary models of generative AI

The debate about proprietary versus open source development models has been going on in the software industry for decades, well illustrated by the example of Microsoft versus the Linux operating system. For some time now, it has also been taking place for generative artificial intelligence models. What are the advantages (+) and what are the disadvantages (-)? Our tech know-how to go explains.

Proprietary models:

  • + customised support and specialist knowledge of the provider
  • + High level of reliability
  • + Existence of service level agreements (SLAs) and technical support
  • + Security standards, especially for business-critical processes
  • + Simple integration into the existing infrastructure
  • - Higher costs, for example due to licence fees
  • - Dependence on one provider

Companies in favour of proprietary models: OpenAI, Microsoft and Google

Open Source:

  • + more transparency
  • + reduced latency
  • + improved because customised performance
  • + significantly greater control over own data through open source code
  • + lower risk of data breaches or unauthorised access
  • + cost efficiency
  • - Misuse possible

Companies favouring open source AI models: IBM, Meta, Intel, Advanced Micro Devices, Oracle, ServiceNow, Sony, SoftBank and Dell Technology.

Interesting links:

https://www.techopedia.com/generative-ai-crossroads-open-source-vs-proprietary-models

https://www.marketwatch.com/articles/ibm-meta-open-ai-models-f33c9110

Related links

Topics

Categories

Contacts

Corporate Communications

Press contact We look forward to your request! +49151 226 75 393

Related media Download images for media use

Supercomputing Center Jülich_wires 2
  • Supercomputing Center Jülich_wires 2
  • License: Media Use
    The content may be downloaded by journalists, bloggers, columnists, creators of public opinion, etc. It can be used and shared in different media channels to convey, narrate, and comment on your press releases, posts, or information, provided that the content is unmodified. The author or creator shall be attributed to the extent and in the manner required by good practice (this means, for example, that photographers should be attributed).
  • File format: .jpg
  • Size: 4480 x 2520, 2.94 MB
Download

Related content

Supercomputer Juwels in Jülich. In autumn 2024 the first exaFlop computer JUPITER will start in Jülich.

Why artificial intelligence needs ExaFlop computers

Exaflop supercomputers for artificial intelligence are needed to provide the computing power required for complex AI algorithms, big data analysis, modeling and simulations, and to enable greater scalability for the increasing demands of AI.

ParTec AG Modular Super and Quantum Computing

ParTec AG specializes in the development and manufacture of modular High-Performance Computing (HPC) systems and Quantum Computers (QC) as well as accompanying system software. Its offering also includes consulting and support services in all areas of the development, construction, and operation of these advanced systems. The dynamic Modular System Architecture (dMSA) paradigm is the result of more than ten years of research and has been developed by ParTec as a novel system design for massively parallel high-computing systems. The dMSA and its enabling ParaStation Modulo Software Suite, developed and maintained by ParTec, have proven particularly successful for the complex requirements of massive computing power in AI. Further information on the company and ParTec AG’s innovative solutions in the field of high-performance computing and quantum computing can be found at www.par-tec.com.

ParTec AG