# granite-timeseries-patchtst-fm-r2 by IBM Granite: Open Model Source: https://savrn.com/models/granite-timeseries-patchtst-fm-r2 Markdown alternate of the page above; the site index is https://savrn.com/llms.txt --- ## Runs On What it takes to serve granite-timeseries-patchtst-fm-r2 (385M parameters): the memory its weights need at each precision, and the cheapest way to rent enough data-center GPUs to hold them. | Precision | Weights | Memory needed | Cheapest setup | Per hour | Also fits | | --- | --- | --- | --- | --- | --- | | 16-bit | 0.8 GB | 0.9 GB | 1x [MI300X](https://savrn.com/ai-index/pricing/gpus/mi300x) (192 GB) Vultr | $1.85 | [1x H100](https://savrn.com/ai-index/pricing/gpus/h100) $1.99 · [1x MI325X](https://savrn.com/ai-index/pricing/gpus/mi325x) $2.00 | | 8-bit | 0.4 GB | 0.5 GB | 1x [MI300X](https://savrn.com/ai-index/pricing/gpus/mi300x) (192 GB) Vultr | $1.85 | [1x H100](https://savrn.com/ai-index/pricing/gpus/h100) $1.99 · [1x MI325X](https://savrn.com/ai-index/pricing/gpus/mi325x) $2.00 | | 4-bit | 0.2 GB | 0.2 GB | 1x [MI300X](https://savrn.com/ai-index/pricing/gpus/mi300x) (192 GB) Vultr | $1.85 | [1x H100](https://savrn.com/ai-index/pricing/gpus/h100) $1.99 · [1x MI325X](https://savrn.com/ai-index/pricing/gpus/mi325x) $2.00 | Memory is the weights at that precision plus 20% for the runtime and a short context; a long context needs more. Prices are the lowest on-demand hourly rates in the [SAVRN Index](https://savrn.com/ai-index/pricing/gpus), read Oct 23, 2026. [granite-timeseries-patchtst-fm-r2 on every accelerator the SAVRN Index prices, at every precision](https://savrn.com/models/granite-timeseries-patchtst-fm-r2/gpus) ## Model Card PatchTST-FM-r2, a state-of-the-art zero-shot time series foundation model, represents a continuation of the well-recognized PatchTST model series, building on the original PatchTST and its zero-shot variant PatchTST-FM-r1. PatchTST-FM-r2 brings architectural enhancements as well as an expanded training base on top of its predecessor PatchTST-FM-r1. As of August 31, 2026 Granite-TimeSeries-PatchTST-FM-r2 is the top performing zero-shot model released under a permissive, commercial-friendly open-source license on the GIFT-Eval benchmark. Granite-TimeSeries-PatchTST-FM-r2 ranks #2 when considering all zero-shot, replicable models (see below for more details). The architectural changes in r2… Excerpt from the card by IBM Granite, licensed openmdw-1.0. ## Configuration Architecture PatchTSTFMForPrediction Stored precision float32 Model type patchtst_fm ## Identity and Version Repository ibm-granite/granite-timeseries-patchtst-fm-r2 Publisher IBM Granite Task Time series forecasting Modality Time series Library Not stated by the source Parameters 385M parameters Languages Not stated by the source Revision b125275f9204d37cbb81fe47b9cf5e08a521e829 First published 2026-08-07 Last updated 2026-09-09 ## Files and Weights 10 files, 1.5 GB in total. The weights are 1 file totalling 1.5 GB in safetensors. Weights1 file · 1.5 GB Configuration1 file · 1.8 KB Documentation2 files · 9.3 KB Other5 files · 281.1 KB Repository1 file · 1.5 KB Every file | File | Type | Size | SHA-256 | | --- | --- | --- | --- | | model.safetensors | Weights | 1.5 GB | 32f7730abd71 | | config.json | Configuration | 1.8 KB | — | | LICENSE | Documentation | 289 B | — | | README.md | Documentation | 9.0 KB | — | | figures/crps_zero.png | Other | 72.2 KB | — | | figures/crps_zero_pretrained.png | Other | 68.2 KB | — | | figures/mase_zero.png | Other | 66.0 KB | — | | figures/mase_zero_pretrained.png | Other | 67.2 KB | — | | model.sig | Other | 7.5 KB | — | | .gitattributes | Repository | 1.5 KB | — | ## License and Download License openmdw-1.0 Access Open weights, no gate Download size 1.5 GB [Download from IBM Granite](https://huggingface.co/ibm-granite/granite-timeseries-patchtst-fm-r2) Released by IBM Granite through its official repository on Hugging Face. ## Built From - Described by arXiv:2005.08100 - Described by [arXiv:2211.14730](https://savrn.com/papers/a-time-series-is-worth-64-words-long-term-forecasting-with-transformers) - Described by [arXiv:2403.07815](https://savrn.com/papers/chronos-learning-the-language-of-time-series) - Described by arXiv:2508.02879 - Described by [arXiv:2602.06909](https://savrn.com/papers/revisiting-the-generic-transformer-deconstructing-a-strong-baseline-for-time-series-founda) ## Memory Requirements | Precision | Weights in memory | | --- | --- | | As published | 1.5 GB | | 16-bit | 0.8 GB | | 8-bit | 0.4 GB | | 4-bit | 0.2 GB | Weights only, from the published parameter count; the key-value cache and runtime add to this. ## Questions About granite-timeseries-patchtst-fm-r2 ### How much GPU memory does granite-timeseries-patchtst-fm-r2 need? About 0.9 GB at 16-bit and 0.2 GB at 4-bit: the weights (385M parameters) plus a working margin. A long context needs more. ### What is the cheapest GPU to run granite-timeseries-patchtst-fm-r2 on? At 16-bit, 1x MI300X from $1.85 an hour; at 4-bit, 1x MI300X from $1.85 an hour, at the lowest on-demand prices the SAVRN Index lists. ### What license is granite-timeseries-patchtst-fm-r2 released under? openmdw-1.0, as its publisher declares it. Read the license text before commercial use. ## Similar Models Model · Time series forecasting ### [MOMENT-1-large](https://savrn.com/models/moment-1-large) [Auton Lab](https://savrn.com/model-publishers/autonlab) MOMENT is a family of foundation models for general-purpose time-series analysis. The models in this family (1) serve as a building block for diverse time-series analysis tasks (e.g., forecasting, classification, anomaly detection, and imputation, etc.), (2) are effective out-of-the-box, i.e., with no (or few) task-specific exemplars (enabling e.g., zero-shot forecasting, few-shot classification, etc.), and (3) are tunable using in-distribution and task-specific data to improve performance. For details on MOMENT models, training data, and experimental results, please refer to the paper MOMENT: A Family of Open Time-series Foundation Models. Recommended Python Version: Python 3.11 (support… Open weights mit 346M parameters transformers [View model](https://savrn.com/models/moment-1-large) Model · Time series forecasting ### [timesfm-3.0-pytorch](https://savrn.com/models/timesfm-3-0-pytorch) [Google](https://savrn.com/model-publishers/google) TimesFM (Time Series Foundation Model) is a pretrained time-series foundation model developed by Google Research for time-series forecasting. This repository contains the official PyTorch weights and configurations for TimesFM 3.0. This model is released under the TimesFM Non-Commercial License v1.0. timesfm-3.0 is pretrained using - GiftEvalPretrain excluding the datasets that overlap with fev-bench - Wikipedia Pageviews, cutoff Nov 2023 (see paper for details). - Google Trends top queries, cutoff EoY 2022 (see paper for details). - Synthetic and augmented data. title={A decoder-only foundation model for time-series forecasting}, author={Das, Abhimanyu and Kong, Weihao and Sen, Rajat and… Open weights other 331M parameters [View model](https://savrn.com/models/timesfm-3-0-pytorch) Model · Time series forecasting ### [Toto-2.0-313m](https://savrn.com/models/toto-2-0-313m) [Datadog](https://savrn.com/model-publishers/datadog) Toto (Time Series Optimized Transformer for Observability) is a family of time series foundation models for multivariate forecasting developed by Datadog. Toto 2.0 is the current generation, featuring u-μP-scaled transformers ranging from 4m to 2.5B parameters, all trained from a single recipe. Forecast quality improves reliably with parameter count across the family. The family sets a new state of the art on three forecasting benchmarks: BOOM, our observability benchmark; GIFT-Eval, the standard general-purpose benchmark; and the recent contamination-resistant TIME benchmark. Inference code is available on GitHub. For more examples, see the Quick Start notebook and GluonTS integration… Open weights apache-2.0 313M parameters pytorch [View model](https://savrn.com/models/toto-2-0-313m) Model · Time series forecasting ### [moirai-1.1-R-large](https://savrn.com/models/moirai-1-1-r-large) [Salesforce AI Research](https://savrn.com/model-publishers/salesforce) This is new updated version of Moirai-1.0-R (https://huggingface.co/Salesforce/moirai-1.0-R-large). The Moirai-1.1-R model achieved significant improvements (~20%) for low-frequency cases like Yearly and Quarterly data in Normalised Mean Absolute Error (NMAE) for 40 datasets on the Monash repository. This release is for research purposes only in support of an academic paper. Our models, datasets, and code are not specifically designed or evaluated for all downstream purposes. We strongly recommend users evaluate and address potential concerns related to accuracy, safety, and fairness before deploying this model. We encourage users to consider the common limitations of AI, comply with… Open weights cc-by-nc-4.0 311M parameters transformers [View model](https://savrn.com/models/moirai-1-1-r-large) Model · Time series forecasting ### [moirai-1.0-R-large](https://savrn.com/models/moirai-1-0-r-large) [Salesforce AI Research](https://savrn.com/model-publishers/salesforce) Moirai, the Masked Encoder-based Universal Time Series Forecasting Transformer is a Large Time Series Model pre-trained on LOTSA data. For more details on the Moirai architecture, training, and results, please refer to the paper. Fig. 1: Overall architecture of Moirai. Visualized is a 3-variate time series, where variates 0 and 1 are target variables (i.e. to be forecasted, and variate 2 is a dynamic covariate (values in forecast horizon known). Based on a patch size of 64, each variate is patchified into 3 tokens. The patch embeddings along with sequence and variate id are fed into the Transformer. The shaded patches represent the forecast horizon to be forecasted, whose corresponding… Open weights cc-by-nc-4.0 311M parameters transformers [View model](https://savrn.com/models/moirai-1-0-r-large) Model · Time series forecasting ### [timesfm-2.0-500m-pytorch](https://savrn.com/models/timesfm-2-0-500m-pytorch) [Google](https://savrn.com/model-publishers/google) TimesFM (Time Series Foundation Model) is a pretrained time-series foundation model developed by Google Research for time-series forecasting. This is not an officially supported Google product. timesfm-2.0-500m is the second open model checkpoint: - It performs univariate time series forecasting for context lengths up to 2048 time points and any horizon lengths, with an optional frequency indicator. Note that it can go even beyond 2048 context even though it was trained with that as the maximum context. - It focuses on point forecasts. We experimentally offer 10 quantile heads but they have not been calibrated after pretraining. - It ideally requires the context to be contiguous (i.e. no… Open weights apache-2.0 499M parameters timesfm [View model](https://savrn.com/models/timesfm-2-0-500m-pytorch) ## IBM Granite [All models and datasets](https://savrn.com/model-publishers/ibm-granite) ## Versions - [b125275f9204](https://savrn.com/models/granite-timeseries-patchtst-fm-r2/versions/b125275f9204) · current 2026-09-18 ## Explore More - [All time series forecasting models](https://savrn.com/models/tasks/time-series-forecasting) - [All models under openmdw-1.0](https://savrn.com/models/licenses/openmdw-1-0) - [Model comparisons](https://savrn.com/models/comparisons) - [The model directory](https://savrn.com/models) - [Open model prices by host](https://savrn.com/ai-index/pricing/open-models) ## Source - Repository metadata, read 2026-10-20. - [IBM Granite website](https://www.ibm.com/granite) - [Hugging Face record](https://huggingface.co/ibm-granite/granite-timeseries-patchtst-fm-r2) - [How the hub is built](https://savrn.com/model-hub/methodology)