Pending benchmarksTracking since Oct 18, 2026
Is Timestral state of the art?Pending benchmarks.Announced Oct 15, 2026; no independent results yet.
Mistral's first family of time-series forecasting models: zero-shot foundation models in 5M, 28M, 123M and 209M sizes, with open weights on Hugging Face. Not an LLM.
Announced facts
The details so far
- Lab
- Mistral
- Announced
- Oct 15, 2026
- API
- Not yet
- Open weights
- Published
- Parameters
- 5M, 28M, 123M and 209M
- Focus
- Time-series forecasting (zero-shot)
- Forecasting
- Univariate and multivariate, with covariates; 21 quantiles; up to 32,768 steps of history and 1,024-step horizons per pass
- Inference
- PyTorch on CPU or GPU
- Weights
- Hugging Face (repos created 2026-10-14)
- License
- Modified MIT
- API
- Not mentioned
Tags marked unsure are not stated on the lab's own pages.
Next steps
What would let us call it
Verdicts use only independent boards we already track. Two consecutive snapshots are required before any crown.
- Publicly announced with a source
- Usable by the public (API or weights)
- Listed on an independent board we track
- Enough results for a rank in a category
- Holds across two snapshots (crowns only)
Self-reported
What the lab claims
Self-reported. Shown for context; not used in any verdict.
Lab claim: Timestral-209M achieves state-of-the-art zero-shot accuracy on GIFT-Eval, TIME and fev-bench (Mistral's model card).
Receipts
Sources
REPOTimestral-209M on Hugging Face (Mistral AI)huggingface.co/mistralai/Timestral-209M · Oct 15, 2026PRESSLaunch-day post on X by a contributorx.com/aaronabentheuer/status/2110809706570916306 · Oct 15, 2026Tracking continues until an independent board lists it.