List your company

Time series forecasting on GIFT-Eval (test)

0.675MASEFedTRL
-395.7382,280.049754,955.83757,631.62525Feb 19, 2026Mar 5, 2026Mar 20, 2026Apr 4, 2026Apr 19, 2026May 4, 2026May 19, 2026

Evaluation results

MethodLinks
FedTRLType=Zero-Shot Models (FL), Zero-shot=true2026.040.675————————0.468.676———————————
TimesFMType=Zero-Shot Models, Zero-shot=true2026.040.68————————0.4658.237———————————
TabPFNType=Zero-Shot Models, Zero-shot=true2026.040.748————————0.488.268———————————
PTST.Type=Supervised Task-Specific Models, Zero-shot=true2026.040.762————————0.49610.052———————————
FFTSType=Zero-Shot Models (FL), Zero-shot=true2026.040.766————————0.5219.787———————————
ChronosType=Zero-Shot Models, Zero-shot=true2026.040.786————————0.55114.309———————————
iTrans.Type=Supervised Task-Specific Models, Zero-shot=true2026.040.802————————0.52411.32———————————
MoiraiType=Zero-Shot Models, Zero-shot=true2026.040.809————————0.51510.175———————————
N-BEATSType=Supervised Task-Specific Models, Zero-shot=true2026.040.842————————0.68921.381———————————
FedAvgType=Zero-Shot Models (FL), Zero-shot=true2026.040.872————————0.718.231———————————
Auto ARIMAType=Statistical Methods, Zero-shot=true2026.040.964————————0.7721.515———————————
Auto ThetaType=Statistical Methods, Zero-shot=true2026.040.978————————1.05124.031———————————
TiDEType=Supervised Task-Specific Models, Zero-shot=true2026.040.98————————0.65218.557———————————
Seasonal NaiveType=Statistical Methods, Zero-shot=true2026.041————————126.175———————————
DeepARType=Supervised Task-Specific Models, Zero-shot=true2026.041.206————————0.72118.938———————————
NaiveType=Statistical Methods, Zero-shot=true2026.041.26————————1.38328.072———————————
Chronos-2Horizon (h)=642026.055.709————————102.4—255,127132156.20.460.3762.4844.1888,0256925.5—
Chronos-BoltHorizon (h)=642026.056.068————————101.9—251,450131.7156.80.5040.4161.5545.2587,68067.222.7—
Chronos-T5Horizon (h)=642026.056.291————————115.2—292,023147.4173.70.4840.3935.8446.21.986519.1—
NanoTS (Full Finetune)Initialization=Language-pretrained (Qwen3-0.6B), Training Regime=Full Finetune, Horizon (h)=64, Training step=1282026.056.424————————119—271,341143.7168.20.5860.4952.8548.3696,208542.6—
NanoTS (IO Only)Initialization=Language-pretrained (Qwen3-0.6B), Training Regime=IO Only, Horizon (h)=64, Training step=1282026.056.502————————141.5—319,226147.9174.90.5570.4622.6850.31.3261.21.5—
NanoTS (LoRA Attn)Initialization=Language-pretrained (Qwen3-0.6B), Training Regime=LoRA Attn, Horizon (h)=64, Training step=1282026.056.602————————137.2—259,652140.81640.5810.4875.3950.52.2755.52.9—
ARIMAHorizon (h)=642026.056.609————————116.6—345,041145.4171.80.5340.4465.847.82.2760.810.5—
NanoTS (LoRA Attn+IO)Initialization=Language-pretrained (Qwen3-0.6B), Training Regime=LoRA Attn+IO, Horizon (h)=64, Training step=1282026.056.628————————138.8—265,5461441690.5660.4732.8248.51.3657.15.2—
NanoTS (IO Only)Initialization=Random, Training Regime=IO Only, Horizon (h)=64, Training step=1282026.057.341————————141.3—445,540183.52050.5630.4781.1751.7394,85065.33.6—
NanoTS (Full Finetune)Initialization=Random, Training Regime=Full Finetune, Horizon (h)=64, Training step=1282026.057.7————————168.9—583,696210.5232.70.5750.494693,45253.4377,35864.20.9—
NanoTS (LoRA Attn)Initialization=Random, Training Regime=LoRA Attn, Horizon (h)=64, Training step=1282026.057.77————————171.4—592,435212.3234.70.5790.499583,78653.6339,88463.90.1—
NanoTS (LoRA Attn+IO)Initialization=Random, Training Regime=LoRA Attn+IO, Horizon (h)=64, Training step=1282026.057.812————————137.9—346,961166.8190.50.7190.62855,01256.9104,02650.23—
Chronos2(Data leakage)Params=120M, Status=Data leakage2026.0269.8—————————————————————
TimesFM-2.5Params=200M, Zero-shot=true2026.0270.5—————————————————————
PatchTST-FMParams=260M, Zero-shot=true2026.0270.7—————————————————————
Xihe-MaxParams=1.5B, Zero-shot=true2026.0271.1—————————————————————
ReversoParams=2.6M, Zero-shot=true2026.0271.1—————————————————————
TiRexParams=30M, Zero-shot=true2026.0271.6—————————————————————
Xihe-BaseParams=700M, Zero-shot=true2026.0271.8—————————————————————
Chronos2Params=120M, Zero-shot=true2026.0272—————————————————————
Xihe-FlashParams=300M, Zero-shot=true2026.0272.6—————————————————————
Reverso-SmallParams=0.6M, Zero-shot=true2026.0272.6—————————————————————
FlowState-9.1MParams=9.1M, Zero-shot=true2026.0272.6—————————————————————
Moirai2Params=11M, Zero-shot=true2026.0272.8—————————————————————
Xihe-LiteParams=94M, Zero-shot=true2026.0272.9—————————————————————
FlowState-2.6MParams=2.6M, Zero-shot=true2026.0273.5—————————————————————
Kairos-50MParams=50M, Zero-shot=true2026.0274.2—————————————————————
Kairos-23MParams=23M, Zero-shot=true2026.0274.8—————————————————————
Toto-BaseParams=150M, Zero-shot=true2026.0275—————————————————————
Sundial-BParams=128M, Zero-shot=true2026.0275—————————————————————
Kairos-10MParams=10M, Zero-shot=true2026.0275.3—————————————————————
TTM-FinetunedParams=1.0M, Training Mode=Finetuned2026.0275.6—————————————————————
TimesFM-2.0Params=500M, Zero-shot=true2026.0275.8—————————————————————
Reverso-NanoParams=0.2M, Zero-shot=true2026.0276—————————————————————
Xihe-TinyParams=9.5M, Zero-shot=true2026.0276.6—————————————————————
TabPFN-TSParams=11M, Zero-shot=true2026.0277.1—————————————————————
YingLong-300MParams=300M, Zero-shot=true2026.0279.8—————————————————————
YingLong-110MParams=100M, Zero-shot=true2026.0280.9—————————————————————
Chronos-Bolt-SParams=48M, Zero-shot=true2026.0282.2—————————————————————
YingLong-50MParams=50M, Zero-shot=true2026.0282.2—————————————————————
SuperLinearParams=2.5M, Zero-shot=true2026.0285.7—————————————————————
Moirai-LParams=311M, Zero-shot=true2026.0287.5—————————————————————
Chronos-BParams=200M, Zero-shot=true2026.0287.6—————————————————————
YingLong-6MParams=6.0M, Zero-shot=true2026.0288—————————————————————
Moirai-BParams=91M, Zero-shot=true2026.0290.1—————————————————————
Moirai-SParams=14M, Zero-shot=true2026.0294.6—————————————————————
Qwen3 TextHorizon (h)=642026.059,911————————489.1—213.1489.1720.85,672824.63.2356.81.6258.94.8—
Auto ArimaModel Category=Statistical Models2026.02—3.57.534.69.678.2510.9310.458.62—————————————
Auto ETSModel Category=Statistical Models2026.02—5.6710.594.811.9312.513.0711.7510.9—————————————
Auto ThetaModel Category=Statistical Models2026.02—4.511.75811.73913.478.1510.52—————————————
CrossformerModel Category=Deep Learning2026.02—169.6211.67.615.7587.759.42—————————————
DeepARModel Category=Deep Learning2026.02—9.839.566.86.9334.48.17.66—————————————
DLinearModel Category=Deep Learning2026.02—11.679.03108.88.759.538.89.23—————————————
Empirical GPModel Category=Statistical Models2026.02—8.677.725.27.738.55.878.57.56—————————————
Gaussian RBF GPModel Category=Statistical Models2026.02—14.8311.9713.810.7310.7512.213.5512.36—————————————
iTransformerModel Category=Deep Learning2026.02—4.174.346.22.932.253.43.353.77—————————————
MoiraiAgentAdditional learning cost=fine-tune 3B LLM, Router size=3B LLM, Inference latency=472.6 ms2026.06—————————————————————68.87
N-BEATSModel Category=Deep Learning2026.02—7.839.199.28.736.57.676.98.22—————————————
NaiveModel Category=Statistical Models2026.02—10.1712.7512.414.5314.7515.2711.6513.09—————————————
PatchTSTModel Category=Deep Learning2026.02—3.333.164.82.731.753.433.13—————————————
Seasonal NaiveModel Category=Statistical Models2026.02—7.57.889.411.9313.2511.279.79.68—————————————
Spectral Mixture GPModel Category=Statistical Models2026.02—13.1710.53149.29.59.212.710.87—————————————
SynapseAdditional learning cost=training-free adaptive ensemble2026.06—————————————————————69.86
TFTModel Category=Deep Learning2026.02—3.834.225.633.252.534.653.87—————————————
TiDEModel Category=Deep Learning2026.02—11.336.169.67.88.255.877.11—————————————
TimeCopilotAdditional learning cost=no training; LLM agent, Router size=LLM-scale, Inference latency=≥ 450 ms2026.06—————————————————————70.51
TimeRouterAdditional learning cost=∼ 110 s CPU, Router size=1.72 MB, Inference latency=9.9 ms2026.06—————————————————————67.65
TSOrchestraAdditional learning cost=fine-tune reasoning LLM (R1-style), Router size=3B LLM, Inference latency=≥ 472.6 ms2026.06—————————————————————67.68
ZooCastAdditional learning cost=train co-embedding selector2026.06—————————————————————69.2