01Forecast mode
Single model
You pick the model. Use this when you already know which one fits your data.
Live status unavailableView status
Forecasting infrastructure · v1
Time-series forecasting through one API.
Send your series, get a forecast back. Pick a model, let Ephemeris pick one, or run an ensemble. You pay only for the models that ran.
Observed seriesForecast horizon
History Median Interval
01 · Forecast modes
One request format. Three ways to run it.
The request stays the same. Only the mode changes.
02Forecast mode
Auto-select
Ephemeris looks at your series and picks the best model for it.
03Forecast mode
Ensemble
Runs several models and blends their forecasts into one, with uncertainty ranges.
02 · Models
Models available now
Live catalog reconnecting
The API remains the source of truth. Check the status page for the latest reading.
03 · Request lifecycle
What happens to each request
- 01
Validate
Shape, limits, and model capabilities are checked first.
- 02
Reserve
The maximum possible cost is reserved before work begins.
- 03
Forecast
Your chosen mode runs.
- 04
Charge
You are charged only for models that returned a result. The cost is included in the response.
04 · Start with one call
One call to get started
Every response lists the models used, their latency, and what it cost.
Explore the complete request schema →cURLPOST /api/v1/forecast
curl https://ephemeris-staging.cascade.industries/api/v1/forecast \
-H "Authorization: Bearer pc_live_your_key" \
-H "Content-Type: application/json" \
-d '{"mode":"route","series":[{"values":[42,47,45,52,58,61]}],"horizon":12}'Your first forecast is one key away