Research Director
Summary Bullets:
• Eight prominent LLM/GenAI competitors are featured in GlobalData’s new LLM Competitive Landscape Assessment. Google Gemini has been named Leader.
• In addition to core model capabilities, other factors are considered, including enterprise tools and partner ecosystems built around the GenAI model.
GlobalData has just released its first LLM Competitive Landscape Assessment, highlighting the strengths and challenges of eight heavy-hitters in this space.
The new report evaluates the competitors’ differentiators of core model technology including context windows, multimodal and multilingual capabilities, vertical and horizontal use cases, AI guardrails, ecosystem, professional services, and go-to-market strategies.
Within this report, Google has been named ‘Leader’ by GlobalData due to a combination of highly developed model capabilities in the Google Gemini family and sophisticated enterprise tooling to build and scale generative AI (GenAI) applications. OpenAI is ‘Very Strong’ thanks to its core model technology with solid code generation and multilingual capabilities, multimodality, and context window size. Microsoft is also rated ‘Very Strong’ for its high degree of penetration in the enterprise and tooling anchored in the powerful capabilities of its exclusive GenAI partner, OpenAI, and for Microsoft’s proprietary model, Phi. IBM is also rated in the ‘Very Strong’ category for its strengths in generating computer code, along with a broad range of native language support and third-party model support. Amazon, Anthropic, and Meta have been ranked ‘Strong,’ and Cohere is ranked ‘Competitive.’
GenAI platforms are largely based on multimodal foundation models and large language models (LLMs); these are borne out of growing interest in accessing natural language processing (NLP) to query computers, following the significant advancements in AI seen in recent years. These include machine learning and deep learning via neural networks, also called generative adversarial networks (GANs), and finally the emergence of the ‘transformer’ architecture in 2017, representing breakthrough efficiencies in training models.
The phenomenon of GenAI builds on the precursor of new software architectures, hybrid cloud, automation, and advancements in AI, resulting in the emergence of LLMs. LLMs are deep learning models trained using vast amounts of text. They are designed to produce new levels of content creation, automation of repetitive tasks, and deliver personalization for improving the customer experience, and their performance depends on the quality and size of the pretraining dataset.
The term LLM was largely unheard of before the end of 2022, which makes its explosive growth and mega investments both startling and somewhat overwhelming. Its immaturity also makes evaluating competitors’ closely guarded training methodologies challenging. The new comprehensive competitive landscape assessment also includes market drivers, buying criteria, and vendor and buyer recommendations, in addition to the eight product evaluations (please see Large Language Models (LLM): Competitive Landscape Assessment.
Following OpenAI’s release of ChatGPT, major cloud and platform providers, hardware/chip manufacturers, and startups moved quickly, recognizing the potential of a revolutionary technology unparalleled since the birth of the internet. Over the past 18 months, these vendors have leveraged their previous AI development efforts and set out building and training models as part of a GenAI portfolio to help address customers’ business transformations.
