Suggestions appear as you type. Use the up and down arrows to choose one and Enter to open it.

Flashcrash review

#14 of 19 in Time Series Intelligence Software

A focused forecasting model for teams that need multivariate probabilistic predictions.

7.7/10Editor score
Flashcrash7.7 Request access

Reviewed by iTechGuides Editors · Editorial team · Updated Oct 2026

Flashcrash is a pretrained model for forecasting recurring quantities such as demand, traffic, sales, load, and prices. It is aimed at mid-market and enterprise teams that want probabilistic deep-learning forecasts without fine-tuning. Its zero-shot approach uses historical series as context, with up to 2,048 past observations, and can forecast several related series together. Teams can also supply known future input series, which gives the model a way to incorporate information about upcoming conditions.

The main distinction is the range of outputs and relationships it handles. Flashcrash produces 21 quantile levels per step, giving forecasts across a range of possible outcomes rather than a single prediction. It also estimates lead-lag relationships across related time series. Longer horizons are generated step by step, so this is a forecasting workflow centered on sequential predictions rather than a broad analytics suite. CPU, Apple Silicon, and NVIDIA GPU execution provide options for running the model in different computing environments.

Deployment is self-hosted, with Linux and macOS named as platforms. Access is currently offered to select partners, and interested organizations are directed to request access. The pricing model is contact sales, so prospective adopters should expect to discuss access and commercial terms directly. Email is the listed support channel. Flashcrash is a fit for teams focused on multivariate time-series prediction and probabilistic outputs that can work within a self-hosted model deployment. Organizations seeking a generally available product, or a wider analytics and integration ecosystem, should consider alternatives suited to those requirements.

Flashcrash pros and cons

  • Where it wins
    • Produces 21 quantile levels per forecast step
    • Forecasts multiple related series and estimates lead-lag relationships
    • Runs on CPU, Apple Silicon, and NVIDIA GPUs
  • Where it doesn't
    • Access is currently offered to select partners
    • Longer forecast horizons are generated step by step
    • Organizations must contact sales about pricing

Flashcrash fact sheet, pricing and score →

Advertiser disclosure: iTechGuides is reader-supported. Vendors can pay for top positions in our rankings and for a place on other products' pages, and we may earn a commission when you click some links. Bought places stand at the number bought; this page and our disclosure say so. Paying vendors may get a fuller, more detailed review from our editors, and that review can change their score. How we rank.

Last updated · How we research and update