Model Comparison

TimesFM vs Moirai: long-horizon forecasting compared

Google TimesFM is a 200M patched-decoder optimized for long-horizon accuracy. Salesforce Moirai is a universal transformer built to handle any variable count, any frequency, and covariates natively. Both are strong — the trade-off is simplicity vs flexibility.

Head-to-head benchmarksLong-horizon comparisonTry both on your data

Compare in playground View all models

Side by sideSame request, different models

# TimesFM
{"model": "google/timesfm-2.0", "inputs": [...], "parameters": {"prediction_length": 28}}

# Moirai
{"model": "salesforce/moirai-1.1-r-base", "inputs": [...], "parameters": {"prediction_length": 28}}

Swap the model ID. Everything else stays the same.

TimesFM

200M patched-decoder, strong long-horizon accuracy, up to 2048 context

Moirai

Universal transformer, native covariates, up to 5000 context

On TSFM.ai

Both available through one API with identical request shapes

Feature comparison

Dimension	TimesFM	Moirai
Architecture	Patched-decoder transformer	Universal forecasting transformer
Parameters	200M (single model)	14M to 311M across Small, Base, Large
Context length	Up to 2048 time steps	Up to 5000 time steps
Covariate support	Univariate primary — limited covariate handling	Native multivariate with past + future covariates
Quantile support	Point forecasts with distribution head	Flexible mixture distribution outputs
Pre-training data	Google Trends + public time series corpora	LOTSA: 27B observations across 9 domains
Long-horizon strength	Strong benchmark results on extended horizons	Competitive on long horizons with covariate advantage
Typical latency	80–150ms	60–250ms depending on size and context

When to choose which

Choose TimesFM when

You want a single well-tuned model without size selection decisions, need strong long-horizon accuracy on univariate series, or prefer the simplicity of a one-model-fits-most approach backed by Google's pre-training corpus.

Choose Moirai when

Your forecasting problem involves covariates or exogenous variables, you need multivariate support, your series require very long context windows beyond 2048 steps, or you work across domains and frequencies that benefit from Moirai's universal design.

Or compare both

Send the same series to both models through the TSFM.ai API. Compare long-horizon accuracy, quantile coverage, and latency on your actual data.

Try in playground

Compare on your own data

Run TimesFM and Moirai on your actual series. Compare accuracy, latency, and quantile coverage before picking a production default.

Open playground Get API key

TimesFM vs Moirai: long-horizon forecasting compared

Feature comparison

When to choose which

Choose TimesFM when

Choose Moirai when

Or compare both

Frequently Asked Questions

Related Pages

Chronos vs TimesFM

Chronos vs Moirai

Best time series foundation models in 2026

Compare on your own data