Retail and Logistics
Demand Forecasting
Time-series foundation models; no named retail customer found.
Named Models
Chronos-2AmazonReleasedOct 2025Runs onsingle datacentre GPU, CPU serverClasstime seriesAmazon (AWS AI) zero-shot forecasting model with covariate in-context learning; report 17 Oct 2025. Context is 8,192 time steps.Commercial use: allowed; Jakarta hosting: unknown; Indonesian support: not stated
TimesFM 2.5GoogleReleasedSep 2025Runs onsingle datacentre GPU, CPU serverClasstime seriesTimesFM version behind BigQuery AI.FORECAST that 3.0 is planned to replace (implied).Commercial use: allowed; Jakarta hosting: unknown; Indonesian support: not statedMQ-RNN and MQ-CNNAmazonReleasedNov 2017Runs onnot in the notesClasstime seriesMulti-horizon quantile recurrent forecaster, demonstrated on forecasting demand for items sold on Amazon. Paper by Wen et al. (2017).Commercial use: allowed; Jakarta hosting: unknown; Indonesian support: not statedWhere this comes from
Models the research names for this use case. Research, with sources on each model record.
Who Runs It
No deployment is recorded for this use case.
Where this comes from
Documented uses by named organisations. The evidence column says who made the claim.
Candidate Models
TSB-AD
A general benchmark for this model class, not a test on this use case. Time-series anomaly detection, 1,070 series from 40 datasets (VUS-PR); NeurIPS 2024; open to community submissions from 1 Apr 2026. Higher is better.
PaAnoSungkyunkwan University vendor-reported0.53(Sub)-PCA independent0.42
MOMENT (MOMENT-1-large)Carnegie Mellon University independent0.39DeepAnT independent0.31OmniAnomaly maintainer-run0.31PCA independent0.3100.51
TimesFM 3.0Google, time seriesReleasedAug 2026open weightshas scoreslast 18 months
Toto 2.0 1BDatadog, time seriesReleasedJun 2026open weightshas scoreslast 18 monthsToto 2.0 2.5BDatadog, time seriesReleasedJun 2026open weightshas scoreslast 18 monthsToto 2.0 313MDatadog, time seriesReleasedJun 2026open weightshas scoreslast 18 months
Granite PatchTST-FM r1IBM, Rensselaer Polytechnic Institute, time seriesReleasedMar 2026open weightshas scoreslast 18 months
Moirai 2.0 smallSalesforce, time seriesReleasedNov 2025open weightshas scoreslast 18 monthsFlowState r1.1IBM, time seriesReleasedSep 2025open weightshas scoreslast 18 months
TiRexNX-AI, time seriesReleasedMay 2025open weightshas scoreslast 18 monthsGranite TSPulseIBM, time seriesReleasedMay 2025open weightshas scoreslast 18 monthsGranite TinyTimeMixer (TTM) r2IBM, time seriesReleased2024open weightsruns on edgehas scoresShowing 10 of 66. All models of this class
Where this comes from
Derived, not researched: models in the directory whose class matches a class this use case uses, excluding family records and models shut down or deprecated. Ranked by how many fit facts they meet, then by release date. Each fit fact is read from the model record.
Links
ModelsTimesFM 2.5Google, time seriesChronos-2Amazon, time seriesMQ-RNN and MQ-CNNAmazon, time series
TimesFM 2.5Google, time seriesChronos-2Amazon, time seriesMQ-RNN and MQ-CNNAmazon, time seriesWho Runs Itnone recorded
Sources
- google/timesfm-2.5-200m-pytorch , Hugging Face huggingface.co https://huggingface.co/google/timesfm-2.5-200m-pytorch
- amazon/chronos-2 , Hugging Face huggingface.co https://huggingface.co/amazon/chronos-2
To cite this page: Project AURION, AI Use-Cases Research & Innovation, "Demand Forecasting", https://aurion.semai.id/usecases/ret-demand-forecasting/, records of October 2026.