Mistral Timestral: Time-Series Foundation Models From 5M to 209M Parameters

Mistral AI has released Timestral, its first family of time-series foundation models, and its largest variant at 209M parameters ranks first on all three of the field's main public benchmarks — GIFT-Eval, TIME, and fev-bench — beating the previous state of the art with only two-thirds of the model parameters. The release matters because it brings the pre-training paradigm that reshaped language and vision to enterprise time-series data, and because the whole family runs on CPU, making zero-shot forecasting practical inside existing monitoring and planning pipelines.

What Timestral is

Timestral is a family of time-series foundation models (TSFMs) that produces probabilistic, multivariate forecasts and supports both past-only and known-future covariates. Mistral describes it as its first family of time-series foundation models.

The models are pre-trained on a large and diverse corpus so that learned patterns transfer across series, allowing zero-shot forecasting on data the model has never seen — including domains where data is scarce. Results have been submitted to the GIFT-Eval and TIME leaderboards.

Benchmark results

Among pre-trained, zero-shot models, Timestral ranks first across all five metrics on the three main public forecasting benchmarks, according to Mistral. These results come from a single checkpoint, without ensembling.

GIFT-Eval: Timestral appears in the top ten of 83 comparable pretrained and zero-shot entries. Lower is better.
TIME: Timestral appears in the top ten of the 35 leaderboard entries. Lower is better on both metrics.
fev-bench: Timestral appears in the top ten of the 37 leaderboard entries, measured by SQL skill score. Higher is better.

Mistral states that the 209M model outperforms the previous state of the art with only two-thirds of the model parameters.

The four model sizes and their accuracy-latency trade-off

Timestral is released in four sizes — 5M, 28M, 123M and 209M parameters — all sharing the same architecture and training recipe. Mistral claims every size sits on the Pareto front of accuracy versus model size: for each model size, no public model forecasts better.

Specific comparisons Mistral reports:

The 28M model beats Chronos-2, which is about four times larger, on GIFT-Eval and TIME.
The 209M model beats TimesFM-3 and Falcon-2.0 with fewer parameters.
The 123M model comes within 0.2% of the 209M score at three quarters of its latency.
The 28M model is five times faster than the 209M model.

CPU inference without a GPU

Timestral does not need a GPU at inference. On a standard server CPU with 8 threads, forecasting one day ahead from three weeks of 15-minute history takes 10 ms with the 5M model and 140 ms with the 209M model. Smaller models trade some accuracy for speed.

Key technical contributions

Mistral attributes Timestral's results to four combined ideas, each of which will be described in detail in a forthcoming technical report.

1. Decoupled Horizon Forecasting (DHF)

Causal decoders trained with contiguous patch masking, as used in Toto 2.0 and TimesFM-3, mask out the values to be forecast, so those values look like missing values and are also lost as context for later forecasts. Mistral's DHF approach separates the two: the full history stays in the sequence, and learned forecast tokens read the history before them. Every observed value therefore serves as both target and context, true gaps stay flagged as missing, and training horizon lengths and missing-data augmentation can be tuned independently of each other.

2. Covariate-aware attention

Each input series is assigned a role: target variate, past-only covariate, or known-future covariate. Known-future covariates are visible across the whole window while targets stay causal, and information flows only from covariates to targets. Timestral can therefore use a holiday calendar or a weather forecast without leaking a target's own future into its forecast.

3. Synthetic-first data

The majority of Timestral's training data was synthetic. Mistral first generated univariate processes with coherent temporal dynamics, including trends, seasonality and diverse noise patterns. It then used a multivariate synthesizer to construct multivariate series by introducing nonlinear dependencies between those processes.

4. Two-stage training

Timestral is first pre-trained at scale on purely synthetic data with a context size of 8192, then post-trained on a mix of real and synthetic data with a 4x longer context. The training objective is a pinball loss over 21 quantiles, a discrete approximation of CRPS.

Why time-series foundation models matter for enterprises

Time series is one of the most common data types in enterprise, from sensor telemetry to observability metrics, and forecasts built on it drive decisions in capacity planning, energy, supply chains, risk and maintenance.

Mistral frames the field's progression in three stages. Classical models such as ARIMA and linear regression remain well-established tools, but they are usually fitted to a single series and learn only from that series' history. Deep learning models such as PatchTST add the capacity to learn nonlinear patterns across many related series, but they are still typically trained from scratch on each dataset, so what they learn does not carry over to new domains. Time-series foundation models bring the pre-training approach to time series, learning patterns that transfer across series and forecasting zero-shot on unseen data.

An industrial example: thousands of sensors in a plant

Mistral illustrates the model's capabilities with an industrial plant containing thousands of sensors across its equipment.

Multivariate forecasting. A pump's vibration, temperature and pressure do not move independently; a rise in one often explains or causes a change in another. Timestral models related series jointly so it can use those dependencies instead of forecasting each sensor in isolation.
Probabilistic forecasting. A single predicted value may not show the full picture. Timestral predicts a full set of quantiles, so teams can see how wide the uncertainty is, set alert thresholds that reflect it, and plan maintenance around risk rather than a single number.
Categorical covariates. Some covariates take values from a set of categories rather than continuous numbers, such as the day of the week or a machine's state (running, idle, under maintenance). Encoding these as numbers would impose an order or scale they do not have. Timestral takes them as categorical inputs directly.
Known-future covariates. Some drivers of sensor behaviour are known in advance, such as production schedules, planned shutdowns and weather forecasts. Timestral can condition on them, so a scheduled load increase is expected rather than flagged as a surprise.

From model to system

Timestral is a numerical model that can run inside existing monitoring pipelines or be paired with Mistral's language models in multi-modal agentic workflows. Mistral sketches this plant workflow:

Timestral runs continuously on the plant's sensor streams, detecting anomalies as they occur.
When it flags an anomaly on a pump, an agent embeds the signals and compares against embeddings of past sensor data to find similar historical failures.
The agent starts sub-agents to run a root cause analysis, drawing on plant context such as maintenance records, shift logs and operator notes, equipment manuals, past work orders, and readings from related equipment.
The agent drafts a suggested diagnosis based on the findings, with supporting evidence and possible next steps.

The plant manager gets a diagnosis and a plan backed by the sensor data, rather than just an alert.

Adapting Timestral with Mistral Forge

Public benchmarks measure general performance, but production data can differ significantly in behaviour and structure, and it may shift over time as markets change regime, equipment degrades and demand patterns move. Mistral argues the lasting value lies in adapting a model to in-domain data and retraining it as the distribution moves.

Timestral is available through Mistral Forge, the same training pipeline Mistral uses for its language models. It supports:

Zero-shot inference out of the box.
Architectural extensions, such as domain-specific covariates.
Fine-tuning on proprietary data, specialising to a domain while preserving general performance.
Control over the training data, including adjusting the training mix, replaying pre-training data, and generating synthetic data samples to reflect the target domain. Mistral states that full control over the data ensures nothing that was backtested was seen during pre-training.

The full pipeline can be deployed on clients' hardware, including air-gapped environments.

Specifications and availability

Attribute	Timestral
Type	Time-series foundation model (TSFM)
Parameters	5M, 28M, 123M and 209M
Context length	32,768 time steps
Output	Probabilistic forecasts, 21 quantiles
Inputs	Multivariate targets, past-only and known-future covariates
Deployment	On-prem, VPC or air-gapped
Licence	Modified MIT

Mistral has published a GitHub repository at github.com/mistralai/timestral so users can run Timestral on their own data. A technical report with the full results will follow. Mistral invites organisations with more specialised forecasting problems, or those interested in research collaboration, to reach out.