NXAI · hosted on Ephemeris
TiRex-2 API
NXAI's xLSTM forecaster: small, fast, and covariate-aware. Its head emits 320 steps per pass; longer horizons are rolled out the way TiRex does it, with the steps already forecast fed back as missing values rather than as forecasts.
Explicit mode
"model": "tirex2"API nametirex2
PublisherNXAI
WeightsNX-AI/TiRex-2
Size38M univariate, 82M multivariate
LicenceApache-2.0
MultivariateYes
CovariatesPast and known-future
Context read2,048 most recent points
Route and ensemble modes use TiRex-2 up to 320 steps. Name it in explicit mode for longer horizons, up to the API's 512-step limit.
01
REST
Call TiRex-2 directly
cURLUTF-8
curl --request POST \
--url https://ephemeris.cascade.industries/api/v1/forecast \
--header "Authorization: Bearer pc_live_your_key" \
--header "Content-Type: application/json" \
--data '{
"mode": "explicit",
"model": "tirex2",
"series": [{ "values": [112, 118, 132, 129, 121, 135, 148, 148], "freq": "M" }],
"horizon": 12,
"quantiles": [0.1, 0.5, 0.9]
}'02
Agents
From Claude, Cursor or any MCP client
Add the Ephemeris MCP server, then ask your agent to forecast with tirex2. The forecast tool takes mode: "explicit", model: "tirex2".
03
Accuracy
How it scores
TiRex-2 is part of the Ephemeris ensemble, which scores better than it on GIFT-Eval, TIME and fev-bench. Full benchmark results.
No GPU to run