time series
Lag-Llama
Early time-series foundation model (USA / Canada affiliations); worse than Seasonal Naive on GIFT-Eval MASE with test leakage flagged.
Levelmodel
Classtime series
Parameters2.45M
Contextnot in the notes
Released12 Oct 2023
Opennessopen weights permissive
Runs onCPU server
Statusactive
Benchmark Results
independent
GIFT-Evaltime series
pretrained, test leakage Yesindependentsaw test data
1.228
00.71.4
Score in MASE (geo-mean, normalised to Seasonal Naive). Lower is better.
Leaderboard rebuilt from the published raw results, 143 entries covering all 97 configurations
pretrained, test leakage Yes; rank is average CRPS rankindependentsaw test data
0.88
00.51
Score in CRPS (geo-mean, normalised to Seasonal Naive). Lower is better.
rank 136 of 143, Leaderboard rebuilt from the published raw results, 143 entries covering all 97 configurations
Where this comes from
Every score names who ran it: vendor-reported, maintainer-run or independent. A rank is shown with the population it was ranked in (gap rule 3).
R&D Topics
Identity
- Developer
- Morgan Stanley, ServiceNow (ServiceNow Research, Montreal), Universite de Montreal, Mila
- Family
- not in the notes
- Released
- 12 Oct 2023 sourcesannouncement page
- Weights published
- 7 Feb 2024
- Record checked
- 10 Oct 2026
Where this comes from
Developer and dates from the research notes. Release date is the vendor's public announcement; weights date is when open weights were published.
Architecture and Size
- Architecture
- dense transformer Decoder-only transformer using lags as covariates; univariate probabilistic
- Total parameters
- 2.45M sources
- Active parameters
- –
- Checkpoint parameters
- –
- Context length
- not in the notes
- Input
- time series
- Output
- time series
Where this comes from
Parameter counts as the developer states them. Where the checkpoint holds a different total, both are shown (register A16).
Openness and Access
- Openness
- open weights permissive
- Licence
- Apache License 2.0 sources
- Access routes
- open weights download
Where this comes from
The licence file governs over registry tags and aggregator labels.
Deployment
- Runs on
- CPU server
- Footprint
- not in the notes
- Runtimes
- not in the notes
Where this comes from
Deployment tiers and runtimes as named in the notes. Footprint is the notes' figure, not a measurement by AURION.
Indonesia
- Indonesian support
- not stated
- Regional languages
- not in the notes
- Jakarta hosting
- unknown
- Languages
- not in the notes
Where this comes from
Indonesia fields are carried on every model; unknown is an allowed value.
Lineage
The notes record no lineage for this model.
Industry Use Cases
No industry use case links this model.
Links
DevelopersMorgan Stanleycompany, USA
ServiceNow (ServiceNow Research, Montreal)company
Universite de Montrealuniversity, Canada
Milanational lab, Canada
Morgan Stanleycompany, USAServiceNow (ServiceNow Research, Montreal)companyUniversite de Montrealuniversity, CanadaMilanational lab, CanadaModelLag-LlamaMorgan Stanley, time series
Lag-LlamaMorgan Stanley, time seriesR&D TopicsTime-Series Foundation Models
Use CasesIndustrial Forecasting (Load, Throughput, Demand)IndustrialDemand ForecastingRetail and LogisticsAnomaly Detection on Sensor, SCADA and Historian DataIndustrialEnergy OptimisationIndustrial+1 moreUse cases of models in the same R&D topic; none is recorded for this model itself.
Who Runs It
PT Amman Mineral Nusa Tenggara (AMMAN)In production, 8 Jul 2026
PT Pertamina Hulu Rokan (PHR)In production, 17 Oct 2025MIND IDResearch, 14 Oct 2025
Shenzhen Energy GroupIn production, 23 Jun 2025+4 moreDeployments of these use cases; none names this model.
PT Amman Mineral Nusa Tenggara (AMMAN)In production, 8 Jul 2026PT Pertamina Hulu Rokan (PHR)In production, 17 Oct 2025MIND IDResearch, 14 Oct 2025Shenzhen Energy GroupIn production, 23 Jun 2025+4 moreDeployments of these use cases; none names this model.Organisations 4
Sources
- Lag-Llama: Towards Foundation Models for Probabilistic Time Series Forecasting arxiv.org https://arxiv.org/html/2310.08278v3
- GIFT Eval - a Hugging Face Space by Salesforce huggingface.co https://huggingface.co/spaces/Salesforce/GIFT-Eval
- huggingface.co/api/models/time-series-foundation-models/Lag-... https://huggingface.co/api/models/time-series-foundation-models/Lag-Llama
To cite this page: Project AURION, AI Use-Cases Research & Innovation, "Lag-Llama", https://aurion.semai.id/models/morgan-stanley-lag-llama/, records of October 2026.