Open Release of the TIRIAS Research LLM Forecast Tool Characterizing Generative AI Models to 2030 including ChatGPT o1 and Beyond

TIRIAS Research has released for public use a new tool designed to track and forecast large language model (LLM) roadmaps. With models like OpenAI's ChatGPT and Google’s Gemini becoming central to the development of digital assistants, autonomous agents, and potentially super intelligent systems, understanding their future trajectories is important. However, much of the detailed information about these models remains undisclosed, making it difficult to gain a clear picture of future LLMs.

While the LLM Forecast Tool is a prototype, it is useful in tracking LLM developments, comparing capabilities, and assessing hardware impacts. Capabilities include:

• Captures historical data and current information from known open-source models to analyze and forecast the unknown attributes of common LLMs.
• Incorporates emerging capabilities, including pre-processing for improved reasoning as we have just seen announced with ChatGPT o1 “Strawberry.”
• Creates roadmap scenarios, compare models, and track progress over time.
• Flexibility to add entirely new attributes as LLMs increase in complexity.

Additionally, it establishes a relationship between model growth and hardware performance, demonstrating how advancements in infrastructure drive LLM performance. Its core strength is its flexibility, allowing users to update or refine assumptions as additional information becomes available.

TIRIAS Research is providing the tool as an open research project as a starting point for collaboration, and we welcome any feedback or contributions that might improve its accuracy or utility. In my work developing AI projects and forecasting cloud infrastructures, I’ve often needed a way to combine insights from hardware trends with AI capabilities. This tool helps capture those insights while providing a common frame or reference for its contributors.

We want to express our thanks to the analysts and researchers whose work and information have been invaluable. We hope this release can contribute to a clearer understanding of LLM advancements and help bring together a collective understanding of the future of AI.

If you are interested in using the tool and and learning more, or contributing to its development, you can request access to the shared Google Sheet and explore the tool further.

LLM Forecast Tool by Tirias Research

Feel free to contact Simon Solotko at [email protected] with questions or suggestions for improving the model and accompanying analysis. Please note that the tool only includes publicly available information about technology roadmaps. All future information is based on forecasts. TIRIAS Research #LLM #OpenAI #Gemini #llama