I like that it's easy to get started with Nixtla's SDK, which makes creating simple forecasts straightforward. It's convenient because I don't have to train my own machine learning models. Review collected by and hosted on G2.com.

Forecasting time series with sparse demand is very hard to do. It's not really possible to cross learn patterns with different weights on the different time series. Unfortunately, the Nixtla client does not provide a way for me to know how big the request is before it is sent. Instead, it just throws exception, and then I have to retry the request with different chunking parameters. Review collected by and hosted on G2.com.

I like the API and the quality of the forecasts we get. I appreciate the sensitivity to a lot of the time series features and the exogenous variables that we put in to see their impacts on our forecasts. Also, the initial setup of Nixtla was super easy, probably the easiest package to ever onboard to. Review collected by and hosted on G2.com.

I think we used to not see a huge impact with fine tuning. So maybe fine tuning could always be improved. Review collected by and hosted on G2.com.

From Nixtla’s ecosystem, I particularly value TimeGPT, which I used in a project focused on predicting kNDVI (Normalized Difference Vegetation Index) across 214,351 time series derived from a geospatial datacube. The model demonstrated adaptability; it effectively captured vegetation dynamics when provided with sufficient historical context.

With fine-tuning and the inclusion of exogenous variables, TimeGPT achieved reliable long-horizon forecasts, outperforming traditional models across multiple metrics while better reproducing the shape, amplitude, and temporal dynamics of the kNDVI signal.

TimeGPT also shows strong potential for gap-filling and temporal interpolation in satellite-derived time series, where missing observations are frequent due to cloud cover or sensor limitations. When given full historical context, it generalizes the underlying temporal patterns remarkably well, making it a valuable tool for Earth observation and remote sensing applications, provided adequate computational resources and contextual information are available. Review collected by and hosted on G2.com.

While TimeGPT is a powerful and well-designed model, adapting it to large-scale geospatial forecasting presented several challenges. The main limitation lies in its lack of native support for geospatial data structures, such as data cubes or spatial indexing systems (e.g., H3, S2). To forecast kNDVI signals, each grid cell had to be manually converted into a string-based unique identifier combining latitude and longitude, and the forecasts had to be executed in multiple batches due to API limits.

TimeGPT also relies on inferred temporal frequencies and predefined exogenous variables, but does not yet provide feature importance or interpretability diagnostics, making it difficult to identify which variables drive the forecasts.

Despite these challenges, these limitations are understandable given the model’s original design for univariate time series and represent opportunities for future development — particularly toward native geospatial compatibility and improved model interpretability. Review collected by and hosted on G2.com.

I appreciate Nixtla for its relatively easy-to-use way of forecasting time series in a zero-shot manner. The API is very straightforward to use, which I really like. The forecasting speed is impressive too, as it works a lot faster than the baseline models I'm comparing it with. The initial setup was quite easy, as the guide provided was helpful. Review collected by and hosted on G2.com.

I would say the only problem I have is the amount of API calls per month. Review collected by and hosted on G2.com.

What I like most is how quickly I can get solid forecasts with minimal setup. The defaults work well, the API is clean, and it handles many time series at once without extra plumbing. Review collected by and hosted on G2.com.

Sometimes, the results are not really great, and I don't know why, so maybe more built-in diagnostics would be really helpful Review collected by and hosted on G2.com.

convenience, efficiency, customisation - oh, and the possibility to forecast in one shot an entire panel of series Review collected by and hosted on G2.com.

not cheap! there are decent free alternatives out there, but lacking the constant upgrades and refinements of course Review collected by and hosted on G2.com.

I like Nixtla's flexibility and prediction power, which are particularly valuable for my analysis and statistical tasks. Review collected by and hosted on G2.com.

I get a sort of difficulty with the API connection, especially connecting it with my graphical user interface of RStudio. I had a problem with the SSH key, but I solved it thanks to the Nixtla team. Also, I guess the initial setup was not that easy. Review collected by and hosted on G2.com.

I primarily use TimeGPT for research purposes, and what I value most is its simplicity and efficiency in enabling high-quality forecasting with minimal setup. The implementation process is seamless, and its integration with Python and Jupyter environments makes it particularly suitable for academic workflows and reproducible research. Review collected by and hosted on G2.com.

I have not encountered any issues — the platform has performed consistently and reliably across all my research applications. Review collected by and hosted on G2.com.

What I really like about Nixtla is how it makes advanced forecasting tools easy to use and teach. For my MBA and EMBA students, that’s a big deal — they can go from basic concepts to real, hands-on forecasting projects without getting lost in technical details.

The Nixtla packages — like StatsForecast, NeuralForecast, and HierarchicalForecast and TimeGPT API — bring together solid research and practical implementation. They run fast, work well with large datasets, and give reliable results right out of the box. This lets me show students not just how forecasting models work, but how to use them in real business contexts.

I also like the open-source spirit behind Nixtla. The documentation is clear, the examples are reproducible, and the team keeps up with the latest ideas in AI and time-series forecasting. It’s become one of my favorite tools for teaching modern forecasting methods in an accessible, hands-on way. Review collected by and hosted on G2.com.

There’s honestly not much to dislike — Nixtla has become one of my go-to tools for teaching. But if I had to point out something, I’d say there is a missing of a Web interface for beginners, especially for MBA students who are new to Python or forecasting concepts. The documentation is solid, but sometimes it assumes a bit of technical background. Review collected by and hosted on G2.com.

Love that this is a home‑grown Mexican project with world‑class engineering.  Its models are impressively easy to use with a pair of lines of code, fast and cost‑efficient, and the catalogue is highly diverse: from classic statistical baselines, through machine‑learning methods, all the way to neural and foundation models like TimeGPT. I have more experience using the statsforecast library, which, if you’ve ever used Dr Rob Hyndman’s revered `forecast` package in R, you’ll feel right at home: the API feels familiar while adding many modern conveniences. Besides these, extras such as a rich suite of error metrics, built‑in cross‑validation, statistical feature generators, scalable execution on both Pandas and PySpark, probabilistic forecast intervals, and even an integrated AI assistant in its webpage to make everyday time‑series work delightfully productive. Review collected by and hosted on G2.com.

Cross‑validation, while powerful, is still hard to configure and not very intuitive. Despite the handy AI helper, clearer in‑line documentation and more usage examples would save time, particularly, when AI hallucinations forces you to double‑check primary sources. Finally, it baffles me that the library isn’t far more popular already; something this good deserves a wider crowd! Review collected by and hosted on G2.com.