TimeGPT does for forecasting models what OpenAI has done for language models – make them accessible to anyone
Time-series forecasting leverages time-stamped data to predict future events
SAN FRANCISCO, July 18, 2024 – Nixtla, a startup that makes it simple for any organization to perform time-series forecasting in seconds, announced today that after a months-long beta, its flagship TimeGPT software is now available to anyone.
Nixtla’s TimeGPT is the first to do for forecasting models what OpenAI has done for language models – making them accessible to all. Nixtla users include Ford, Walmart, FedEx, and Databricks.
Time-series forecasting is a powerful method that leverages time-stamped data to predict future events and remove uncertainty from business conditions – for example, to accurately predict sales, inventory levels and even manufacturing data. Until now, time-series forecasting has required large teams of machine learning engineers developing custom models – which meant only the largest companies could leverage it. With TimeGPT, Nixtla makes time-series forecasting available to anyone with just three lines of code.
“We’re very excited to release this stable version of TimeGPT for all, and grateful for the thousands of beta users who provided us with critical feedback that helped us improve it,” said Max Mergenthaler, co-founder and CEO of Nixtla. “TimeGPT users have explored use cases across finance, retail, manufacturing, healthcare, security and more. We saw a wide range of company types, from very small mom-and-pop businesses to the Fortune 500 – proving that any organization can benefit from time-series forecasting.”
Feedback from one user: “Before TimeGPT, our team spent a lot of time creating and maintaining forecasting pipelines. Now, we do state-of-the-art forecasting in a few lines of code and in just a couple of seconds. TimeGPT saved us major hours and headaches.”
Nixtla recently added several new TimeGPT features including:
- An R package to enable advanced users to access the TimeGPT API from R as well as Python
- Speed and accuracy improvements shown in benchmark testing that makes Nixtla more accurate and twice as fast as other options
- TimeGPT for Excel (Beta): forecast and perform anomaly detection in a few clicks in your Excel spreadsheets. View the demo.
In addition, Nixtla recently announced an integration with Microsoft that brings an optimized version of TimeGPT called TimeGEN-1 to the Azure AI model catalog. It is offered as a Model as a Service (MaaS). This integration means millions of Azure users now have access to Nixtla’s time-series forecasting models.
Nixtla was founded by Mergenthaler, Azul Garza Ramirez (CTO) and Cristian Challu (CSO). Mergenthaler is a serial entrepreneur and engineer with deep experience in building data science teams. Ramirez studied applied math and economics and is an experienced machine learning engineer. Challu has a PhD in machine learning. The three met at a previous startup and decided to build an open-source time series model. When the model’s popularity soared (it has been downloaded 15M times and is in production at multiple Fortune 500 companies), they founded Nixtla to continue development and create an enterprise version. Nixtla means “time” in Aztec, an ancient Mexican language (all three founders, now based in San Francisco, are originally from Mexico).
Other than Nixtla, all other options for AI-driven time-series forecasting come from tech giants: Google, Amazon, Salesforce and ServiceNow.
Get started with TimeGPT
- Go to dashboard.nixtla.io/freetrial to set up an account
- Follow our TimeGPT Quickstart guide
- Read through our tutorials and capabilities, for examples of different ways to use TimeGPT.
