Time Series Forecasting on GIFT-Eval (aggregate)
0.69Relative MASETimer-S1
Evaluation results
| Method | Links | ||
|---|---|---|---|
| Timer-S1Type=Foundation (MoE), Params=8.3B, GPU=Yes, Evaluation Protocol=zero-shot2026.05 | 0.69 | 0.49 | |
| Chronos-2Type=Foundation, Params=205M, GPU=Yes, Evaluation Protocol=zero-shot2026.05 | 0.698 | 0.485 | |
| TimesFM-2.5Type=Foundation, Params=200M, GPU=Yes, Evaluation Protocol=zero-shot2026.05 | 0.705 | 0.49 | |
| TTM-R3Type=Foundation, Params=∼1M, GPU=Yes, Evaluation Protocol=fine-tuned per dataset2026.05 | 0.713 | 0.511 | |
| TTM-R3Type=Foundation, Params=∼1M, GPU=Yes, Evaluation Protocol=zero-shot2026.05 | 0.724 | 0.52 | |
| Moirai2Type=Foundation, Params=311M, GPU=Yes, Evaluation Protocol=zero-shot2026.05 | 0.728 | 0.516 | |
| Chronos-Bolt-BaseType=Foundation, Params=205M, GPU=Yes, Evaluation Protocol=zero-shot2026.05 | 0.808 | 0.574 | |
| FLAIRType=Statistical, Params=P+p+1*, GPU=No, Evaluation Protocol=fit per-series2026.05 | 0.838 | 0.587 | |
| PatchTSTType=Deep Learning, Params=∼1M, GPU=Yes, Evaluation Protocol=per-dataset2026.05 | 0.849 | 0.587 | |
| Chronos-LargeType=Foundation, Params=710M, GPU=Yes, Evaluation Protocol=zero-shot2026.05 | 0.87 | 0.647 | |
| Moirai-LargeType=Foundation, Params=311M, GPU=Yes, Evaluation Protocol=zero-shot2026.05 | 0.875 | 0.599 | |
| Chronos-smallType=Foundation, Params=46M, GPU=Yes, Evaluation Protocol=zero-shot2026.05 | 0.892 | 0.663 | |
| iTransformerType=Deep Learning, Params=∼5M, GPU=Yes2026.05 | 0.893 | 0.62 | |
| TFTType=Deep Learning, Params=∼10M, GPU=Yes2026.05 | 0.915 | 0.605 | |
| N-BEATSType=Deep Learning, Params=∼10M, GPU=Yes2026.05 | 0.938 | 0.816 | |
| Seasonal NaiveType=Baseline, Params=0, GPU=No2026.05 | 1 | 1 | |
| DLinearType=Deep Learning, Params=∼0.1M, GPU=Yes, Evaluation Protocol=per-dataset2026.05 | 1.061 | 0.846 | |
| ETS(A,N,M)Type=Statistical, Params=∼5, GPU=No, Evaluation Protocol=fit per-series2026.05 | 1.067 | 0.833 | |
| AutoARIMAType=Statistical, Params=∼5, GPU=No, Evaluation Protocol=fit per-series2026.05 | 1.074 | 0.912 | |
| AutoThetaType=Statistical, Params=∼5, GPU=No, Evaluation Protocol=fit per-series2026.05 | 1.09 | 1.244 | |
| ETS(M,Ad,M)Type=Statistical, Params=∼5, GPU=No, Evaluation Protocol=fit per-series2026.05 | 1.254 | 1.088 | |
| DeepARType=Deep Learning, Params=∼10M, GPU=Yes2026.05 | 1.343 | 0.853 | |
| ProphetType=Statistical, Params=∼20, GPU=No, Evaluation Protocol=fit per-series2026.05 | 1.54 | 1.061 |