● Covered by 1 source · 1 reportMedium impact1 neutral
IBM releases Granite Time Series PatchTST-FM-r2 model with commercial-friendly license
Key points
- IBM released Granite Time Series PatchTST-FM-r2 model.
- It has ~385M parameters and supports probabilistic forecasting.
- The model is licensed under Apache 2.0 and OpenMDW 1.0.
- It ranks #1 in zero-shot performance on GIFT-Eval benchmark.
New Model Release
IBM has released Granite Time Series PatchTST-FM-r2, the latest iteration in its Granite TSFM family of models. This new version updates its predecessor, PatchTST-FM-r1, by incorporating an updated architecture, a larger pretraining corpus, and new features such as probabilistic forecasting and support for imputation of missing values.
Performance and Capabilities
The PatchTST-FM-r2 model contains approximately 385 million parameters and is designed for general-purpose zero-shot forecasting across various domains like demand, prices, energy loads, and traffic. It offers flexible forecast lengths and probabilistic forecasts through a 99-quantile prediction head. The model's backbone uses conformer blocks that combine multi-head self-attention with temporal convolution to capture both long- and short-range temporal structures.
Benchmarking Results
As of September 8, 2026, Granite Time Series PatchTST-FM-r2 is the top-performing zero-shot model released under a permissive, commercial-friendly open-source license on the GIFT-Eval leaderboard. GIFT-Eval is a comprehensive benchmark for time series forecasting. The model ranks second overall among all replicable, zero-shot models on this benchmark.
Licensing and Availability
The model is dual-licensed under Apache 2.0 and OpenMDW 1.0, allowing users to select either license for commercial-friendly use. IBM has made the model weights, architecture, inference pipeline, and code necessary to reproduce the benchmark results publicly available. The model is accessible on Hugging Face.
✨ This summary was generated by AI from the outlets' reporting listed below. It is not independently verified and may contain errors — check the original sources. How BrevFeed works →
The daily brief
One email each morning: the day's tech stories, clustered across outlets and summarized. No account needed.
One email a day. Unsubscribe in one click, any time.
Today's brief
Spend a few minutes, get the whole day. Every topic's top stories in one hands-free rundown — listen, watch, or read the transcript.
▶ Play today's brief
New every morning, and the back catalogue is archived by date.
Reporting from
IBM has released Granite Time Series PatchTST-FM-r2, an updated time-series foundation model with approximately 385 million parameters, under Apache 2.0 and OpenMDW 1.0 licenses. This model achieves top zero-shot performance among permissively licensed, replicable models on the GIFT-Eval benchmark, offering capabilities like probabilistic forecasting and missing value imputation.