Models · Updated 19 Oct, 03:29 pm IST
Mistral releases Timestral time-series foundation models
Why it matters for readers: If you follow AI models, this shows compact models can match or beat larger ones on forecasting benchmarks.
- Timestral comes in four sizes from 5M to 209M parameters and aims to balance accuracy with efficiency.1
- Mistral says the 209M-parameter model ranks first on the GIFT-Eval, TIME and fev-bench public benchmarks.1
- The models deliver probabilistic, multivariate forecasts, support past-only and known-future covariates, and run efficiently on CPUs.1
- Timestral's training combines synthetic pre-training, covariate-aware attention and long-context learning, and can be fine-tuned via Mistral Forge for domain adaptation.1
Get a brief like this every morning
Uzha reads hundreds of sources and gives you the stories that matter for your work, with every source linked. Free.
Get started