Releases·Model Release
IBM releases Granite Time Series PatchTST-FM-r2, top permissively-licensed zero-shot model on GIFT-Eval
IBM released Granite Time Series PatchTST-FM-r2, the latest model in its Granite TSFM family, under a commercial-friendly license. The model supports zero-shot forecasting, letting users apply a pretrained model to new datasets instead of training per-dataset models. This continues the shift toward foundation-model-based forecasting systems.
0 engagement1 sourceWed, Sep 9, 2026, 03:36 PM
AI Overview
- IBM's ~385M-parameter PatchTST-FM-r2 tops permissive-license zero-shot models on GIFT-Eval (CRPS 0.467, MASE 0.6846), ranking #2 overall as of September 8, 2026.
- PatchTST-FM-r2 uses conformer blocks pairing multi-head self-attention with temporal convolution, 50% overlapping Hamming-windowed patches, and supports 8,192-step contexts with 99-quantile probabilistic forecasts.
- The model is dual-licensed under Apache 2.0 and OpenMDW 1.0, with open weights, inference pipeline, and reproducibility code on Hugging Face.
- IBM and Confluent offer Granite Time Series models (PatchTST-FM-r1, FlowState-r1.1, TTM-r3, TSPulse) via Early Access in Confluent Cloud for streaming inference through Apache Flink.
Related
Products·launched
Ephemeris: API for ensemble forecasts from multiple open-source time series foundation models
Ephemeris is an inference provider exposing several open-source time series foundation models (Chronos2, Flowstate, TimesFM, and others) through one API. Users can select multiple models to produce ensemble forecasts, and can hand the API to an LLM agent to use if unfamiliar with TSFMs. It lowers the barrier to running these models for forecasting without downloading them locally.
Ephemeris is an inference provider exposing several open-source time series foundation models (Chronos2, Flowstate, TimesFM, and others) through one API. Users can select multiple models to produce ensemble forecasts, and can hand the API to an LLM agent to use if unfamiliar with TSFMs. It lowers the barrier to running these models for forecasting without downloading them locally.
1 engagement1 sourceredditFri, Sep 25, 2026, 10:04 AM
Releases·Model Release
$t_0$: Open-Weights Time-Series Foundation Models for Forecasting with Covariate Context
A new arXiv report introduces $t_0$, a family of open-weights time-series foundation models that condition forecasts on past covariates and known-future covariates without task-specific retraining. The first two releases, t0-alpha (102M) and t0-beta (256M), rank third on GIFT-Eval among zero-shot TSFMs and perform strongly on fev-bench and an independent Macrocosm evaluation.
A new arXiv report introduces $t_0$, a family of open-weights time-series foundation models that condition forecasts on past covariates and known-future covariates without task-specific retraining. The first two releases, t0-alpha (102M) and t0-beta (256M), rank third on GIFT-Eval among zero-shot TSFMs and perform strongly on fev-bench and an independent Macrocosm evaluation.
0 engagement1 sourcearxivMon, Sep 21, 2026, 01:22 PM
Releases·Model Release
Show HN: T0: We trained a 256M SOTA open-weights time-series forecasting model
The Forecasting Company launched t0-gamma, a 256M-parameter open-weights foundation model for zero-shot time-series forecasting. It ranks first among time-series foundation models on GIFT-Eval, fev-bench and TIME, and runs on CPU, GPU or Apple silicon. Weights are on Hugging Face, with a pip-installable package, Colab notebook, and arXiv tech report.
The Forecasting Company launched t0-gamma, a 256M-parameter open-weights foundation model for zero-shot time-series forecasting. It ranks first among time-series foundation models on GIFT-Eval, fev-bench and TIME, and runs on CPU, GPU or Apple silicon. Weights are on Hugging Face, with a pip-installable package, Colab notebook, and arXiv tech report.
10 engagement1 sourcehackernewsWed, Oct 14, 2026, 02:51 PM
Social·Community
TimesFM-3: Bringing LLM-Style Foundation Models to Time-Series Forecasting
Medium article examines TimesFM-3's LLM-style approach to time-series forecasting
Medium article examines TimesFM-3's LLM-style approach to time-series forecasting
0 engagement1 sourcerssWed, Sep 2, 2026, 07:35 AM
Releases·Model Release
Google Research announces TimesFM-3, a zero-shot foundation model for multivariate forecasting
Google Research announced TimesFM-3, a zero-shot foundation model for multivariate time-series forecasting. The announcement post appeared via research.google, shared on Reddit. Practitioners in forecasting may find a pretrained zero-shot alternative to task-specific model training.
Google Research announced TimesFM-3, a zero-shot foundation model for multivariate time-series forecasting. The announcement post appeared via research.google, shared on Reddit. Practitioners in forecasting may find a pretrained zero-shot alternative to task-specific model training.
95 engagement1 sourceredditTue, Sep 1, 2026, 12:56 AM