Mistral AI says Timestral leads time-series forecasting benchmarks
🎧 Listen to this story
Mistral AI has released Timestral, its first family of time-series foundation models for probabilistic and multivariate forecasting. The family includes four models with 5 million, 28 million, 123 million and 209 million parameters, supports past and known-future covariates, and can forecast zero-shot on data from new domains.
The company says the 209-million-parameter model ranked first on the field’s three main public benchmarks — GIFT-Eval, TIME and fev-bench — using a single checkpoint and outperforming larger models. According to Mistral AI, Timestral does not need a GPU for inference: on a standard server CPU with eight threads, a one-day forecast based on three weeks of 15-minute history takes 10 ms with the 5-million-parameter model and 140 ms with the 209-million-parameter version. The family can also be fine-tuned for specific domains through Mistral Forge.
Why it matters · editorial interpretation
Mistral AI’s claim of leading three benchmarks and running inference on CPUs suggests that specialized time-series models could combine competitive performance with lower operating costs. Support for fine-tuning through Mistral Forge may also make adaptation to specific domains easier.
Sources
Research